How-To Tutorials - Data

For many years, Elastic Stack has served as an open-source, simple yet powerful interface for security analysts to detect and mitigate malicious behavior. However, Elastic marked its official entry into the security analytics market with Elastic SIEM in June this year. Since its initial release, Elastic SIEM has seen a number of enhancements including machine learning-based anomaly detection, maps integration, and more. To further expand its presence in the security field, Elastic in early October, completed the acquisition of Endgame, a security company focused on endpoint prevention, detection, and response. Following this acquisition, Elastic introduced the Elastic Endpoint Security solution in October to help organizations “automatically and flexibly respond to threats in real-time.” The company has also eliminated per-endpoint pricing. In this article, we will look at what is Elastic SIEM, how it fits into the Elastic Stack, its components, and how a security operations team leverages Elastic SIEM to defend its data and infrastructure against attacks. [box type="shadow" align="" class="" width=""] Further learning This is a quick overview of the Elastic Stack. To learn more check out our book, Learning Elastic Stack 7.0 - Second Edition by Pranav Shukla and Sharath Kumar M N. This book will give you a fundamental understanding of what the stack is all about, and help you use it efficiently to build powerful real-time data processing. [/box] Introducing Elastic SIEM Elastic SIEM is not a standalone product but rather builds on the existing Elastic Stack capabilities used for security analytics including search, visualizations, dashboards, alerting, machine learning features, and more. The following diagram shows how Elastic SIEM fits into the Elastic Stack: Source: Elastic The beta version of Elastic SIEM was released in June this year with Elastic Stack 7.2. It includes a new set of data integrations for security use cases and a dedicated app in Kibana. It enables users to analyze host-related and network-related security events as part of alert investigations, threat hunting, initial investigations, and triaging of events. You can access Elastic SIEM through the Elastic Cloud or by downloading its default distribution. Elastic SIEM supports the recently introduced Elastic Common Schema (ECS), a uniform way to represent data across different sources. ECS defines a common set of fields and objects to ingest data into Elasticsearch enabling users to centrally analyze information like logs, flows, and contextual data from across environments. Features of Elastic SIEM Host-related security event analysis The Hosts view shows key metrics regarding host-related security events and a set of data tables that enable interaction with the Timeline Event Viewer. For further investigation, you can drag-and-drop items of interest from the Hosts view tables to Timeline. This gives you deeper insight into hosts, unique IPs, user authentications, uncommon processes, and events. We can filter the host view with the search bar at the top. To help you search faster, SIEM provides a search experience that combines traditional text-based search with the visual query builder that’s deeply integrated with drag-and-drop throughout the SIEM app and powered by the Elastic common schema. Network-related security event analysis The Network view provides analysts the key network activity metrics and event tables. You can drag-and-drop these tables to Timeline for further investigation to get deeper insight into the source and destination IP, top DNS domains, users, transport layer security certs, and more. Starting with Elastic Stack 7.4, you have Elastic Maps integrated right into Elastic SIEM. The interactive map is created based on live data that analysts can search, filter, and explore in real-time. The map gives analysts an overview of the network traffic. They can simply hover over source and destination points to uncover more details such as hostnames and IP addresses. They can also click a hostname to go to the SIEM Host view or an IP address to open the relevant network details. This integration lets Elastic SIEM leverage geospatial analytics and search capabilities of Elastic Maps. It also uses the new point-to-point line feature to easily visualize the connections in your data. Timeline Event Viewer The Timeline Event Viewer enables security analysts to gather and store evidence of an attack. They can pin and annotate relevant events, comment on and share their findings, and do everything within Kibana. It is a collaborative workspace for investigations or threat hunting where analysts can easily drag objects of interest from Network and Hosts view for further investigation. Anomaly detection with machine learning integration Cyber attacks today have become so sophisticated that it is hard to maintain an effective defense with just a set of static rules. Looking at the importance of automated analysis and detection, Elastic integrated machine learning capabilities right into the SIEM app in 7.3. This allowed security analysts to enable and run a set of machine learning anomaly detection jobs designed to detect specific cyber attack behaviors. The detected anomalies are then displayed on the Hosts and Network views in the SIEM app. However, in Elastic SIEM 7.3, there were only three built-in anomaly detection jobs. In the latest release (7.4), Elastic has added thirteen more anomaly detection jobs some of which are anomalous network activity, anomalous process, anomalous path activity, anomalous Powershell script, and more. This machine learning integration is extensible allowing users to add their own jobs to the SIEM job group. These were some of the key features in Elastic SIEM. Check out the Elastic SIEM 7.4 release announcement to know more. Also, to get a better understanding of how Elastic SIEM works, see the webinar Hands-on with Elastic SIEM: Defending your organization with the Elastic Stack by Elastic. To get started with Elastic Stack you can check out our book Learning Elastic Stack 7.0 - Second Edition. This book will help you learn how to use Elasticsearch for distributed searching and analytics, Logstash for logging, and Kibana for data visualization.  As you work through the book, you will discover the technique of creating custom plugins using Kibana and Beats. The book also touches upon Elastic X-Pack, a useful extension for effective security and monitoring.  You’ll also find helpful tips on how to use Elastic Cloud and deploy Elastic Stack in production environments. How to push Docker images to AWS’ Elastic Container Registry(ECR) [Tutorial] Core security features of Elastic Stack are now free! Elastic Stack 6.7 releases with Elastic Maps, Elastic Update and much more!

If we talk about recent breakthroughs in the software community, machine learning and deep learning is a major contender - the usage, adoption, and experimentation of deep learning has exponentially increased. Especially in the areas of computer vision, speech, natural language processing and understanding, deep learning has made unprecedented progress. GANs, variational autoencoders and deep reinforcement learning are also creating impressive AI results. To know more about the progress of deep learning, we interviewed Ivan Vasilev, a machine learning engineer and researcher based in Bulgaria. Ivan is also the author of the book Advanced Deep Learning with Python. In this book, he teaches advanced deep learning topics like attention mechanism, meta-learning, graph neural networks, memory augmented neural networks, and more using the Python ecosystem. In this interview, he shares his experiences working on this book, compares TensorFlow and PyTorch, as well as talks about computer vision, NLP, and GANs. On why he chose Computer Vision and NLP as two major focus areas of his book Computer Vision and Natural Language processing are two popular areas where a number of developments are ongoing. In his book, Advanced Deep Learning with Python, Ivan delves deep into these two broad application areas. “One of the reasons I emphasized computer vision and NLP”, he clarifies, “is that these fields have a broad range of real-world commercial applications, which makes them interesting for a large number of people.” The other reason for focusing on Computer Vision, he says “is because of the natural (or human-driven if you wish) progress of deep learning. One of the first modern breakthroughs was in 2012, when a solution based on convolutional network won the ImageNet competition of that year with a large margin compared to any previous algorithms. Thanks in part to this impressive result, the interest in the field was renewed and brought many other advances including solving complex tasks like object detection and new generative models like generative adversarial networks. In parallel, the NLP domain saw its own wave of innovation with things like word vector embeddings and the attention mechanism.” On the ongoing battle between TensorFlow and PyTorch There are two popular machine learning frameworks that are currently at par - TensorFlow and PyTorch (Both had new releases in the past month, TensorFlow 2.0 and PyTorch 1.3). There is an ongoing debate that pitches TensorFlow and PyTorch as rivaling tech and communities. Ivan does not think there is a clear winner between the two libraries and this is why he has included them both in the book. He explains, “On the one hand, it seems that the API of PyTorch is more streamlined and the library is more popular with the academic community. On the other hand, TensorFlow seems to have better cloud support and enterprise features. In any case, developers will only benefit from the competition. For example, PyTorch has demonstrated the importance of eager execution and TensorFlow 2.0 now has much better support for eager execution to the point that it is enabled by default. In the past, TensorFlow had internal competing APIs, whereas now Keras is promoted as its main high-level API. On the other hand, PyTorch 1.3 has introduced experimental support for iOS and Android devices and quantization (computation operations with reduced precision for increased efficiency).” Using Machine Learning in the stock trading process can make markets more efficient Ivan discusses his venture into the field of financial machine learning, being the author of an ML-oriented event-based algorithmic trading library. However, financial machine learning (and stock price prediction in particular) is usually not in the focus of mainstream deep learning research. “One reason”, Ivan states, “is that the field isn’t as appealing as, say, computer vision or NLP. At first glance, it might even appear gimmicky to predict stock prices.” He adds, “Another reason is that quality training data isn’t freely available and can be quite expensive to obtain. Even if you have such data, pre-processing it in an ML-friendly way is not a straightforward process, because the noise-to-signal ratio is a lot higher compared to images or text. Additionally, the data itself could have huge volume.” “However”, he counters, “using ML in finance could have benefits, besides the obvious (getting rich by trading stocks). The participation of ML algorithms in the stock trading process can make the markets more efficient. This efficiency will make it harder for market imbalances to stay unnoticed for long periods of time. Such imbalances will be corrected early, thus preventing painful market corrections, which could otherwise lead to economic recessions.” GANs can be used for nefarious purposes, but that doesn’t warrant discarding them Ivan has also given a special emphasis to Generative adversarial networks in his book. Although extremely useful, in recent times GANs have been used to generate high-dimensional fake data that look very convincing. Many researchers and developers have raised concerns about the negative repercussions of using GANs and wondered if it is even possible to prevent and counter its misuse/abuse. Ivan acknowledges that GANs may have unintended outcomes but that shouldn’t be the sole reason to discard them. He says, “Besides great entertainment value, GANs have some very useful applications and could help us better understand the inner workings of neural networks. But as you mentioned, they can be used for nefarious purposes as well. Still, we shouldn’t discard GANs (or any algorithm with similar purpose) because of this. If only because the bad actors won’t discard them. I think the solution to this problem lies beyond the realm of deep learning. We should strive to educate the public on the possible adverse effects of these algorithms, but also to their benefits. In this way we can raise the awareness of machine learning and spark an honest debate about its role in our society.” Machine learning can have both intentional and unintentional harmful effects Awareness and Ethics go in parallel. Ethics is one of the most important topics to emerge in machine learning and artificial intelligence over the last year. Ivan agrees that the ethics and algorithmic bias in machine learning are of extreme importance. He says, “We can view the potential harmful effects of machine learning as either intentional and unintentional. For example, the bad actors I mentioned when we discussed GANs fall into the intentional category. We can limit their influence by striving to keep the cutting edge of ML research publicly available, thus denying them any unfair advantage of potentially better algorithms. Fortunately, this is largely the case now and hopefully will remain that way in the future. “ “I don't think algorithmic bias is necessarily intentional,'' he says. “Instead, I believe that it is the result of the underlying injustices in our society, which creep into ML through either skewed training datasets or unconscious bias of the researchers. Although the bias might not be intentional, we still have a responsibility to put a conscious effort to eliminate it.” Challenges in the Machine learning ecosystem “The field of ML exploded (in a good sense) a few years ago,'' says Ivan, “thanks to a combination of algorithmic and computer hardware advances. Since then, the researches have introduced new smarter and more elegant deep learning algorithms. But history has shown that AI can generate such a great hype that even the impressive achievements of the last few years could fall short of the expectations of the general public.” “So, in a broader sense, the challenge in front of ML is to sustain the current pace of innovation. In particular, current deep learning algorithms fall short in some key intelligence areas, where humans excel. For example, neural networks have a hard time learning multiple unrelated tasks. They also tend to perform better when working with unstructured data (like images), compared to structured data (like graphs).” “Another issue is that neural networks sometimes struggle to remember long-distance dependencies in sequential data. Solving these problems might require new fundamental breakthroughs, and it’s hard to give an estimation of such one time events. But even at the current level, ML can fundamentally change our society (hopefully for the better). For instance, in the next 5 to 10 years, we can see the widespread introduction of fully autonomous vehicles, which have the potential to transform our lives.” This is just a snapshot of some of the important focus areas in the deep learning ecosystem. You can check out more of Ivan’s work in his book Advanced Deep Learning with Python. In this book you will investigate and train CNN models with GPU accelerated libraries like TensorFlow and PyTorch. You will also apply deep neural networks to state-of-the-art domains like computer vision problems, NLP, GANs, and more. Author Bio Ivan Vasilev started working on the first open source Java Deep Learning library with GPU support in 2013. The library was acquired by a German company, where he continued its development. He has also worked as a machine learning engineer and researcher in the area of medical image classification and segmentation with deep neural networks. Since 2017 he has focused on financial machine learning. He is working on a Python based platform, which provides the infrastructure to rapidly experiment with different ML algorithms for algorithmic trading. You can find him on Linkedin and GitHub. Kaggle’s Rachel Tatman on what to do when applying deep learning is overkill  Brad Miro talks TensorFlow 2.0 features and how Google is using it internally François Chollet, creator of Keras on TensorFlow 2.0 and Keras integration, tricky design decisions in deep learning and more

No discussion on image processing can be complete without talking about OpenCV. Its 2500+ algorithms, extensive documentation and sample code are considered world-class for exploring real-time computer vision. OpenCV supports a wide variety of programming languages such as C++, Python, Java, etc., and is also available on different platforms including Windows, Linux, OS X, Android, and iOS. OpenCV-Python, the Python API for OpenCV is one of the most popular libraries used to solve computer vision problems. It combines the best qualities of OpenCV, C++ API, and the Python language. The OpenCV-Python library uses Numpy, which is a highly optimized library for numerical operations with a MATLAB-style syntax. This makes it easier to integrate the Python API with other libraries that use Numpy such as SciPy and Matplotlib. This is the reason why it is used by many developers to execute different computer vision experiments. Want to know more about OpenCV with Python? [box type="shadow" align="" class="" width=""]If you are interested in developing your computer vision skills, you should definitely master the algorithms in OpenCV 4 and Python explained in our book ‘Mastering OpenCV 4 with Python’ written by Alberto Fernández Villán. This book will help you build complete projects in relation to image processing, motion detection, image segmentation, and many other tasks by exploring the deep learning Python libraries and also by learning the OpenCV deep learning capabilities.[/box] At the PyData Warsaw 2018 conference, Sylwek Brzęczkowski walked through how to implement a face swap using OpenCV and Python. Face swaps are used by apps like Snapchat to dispense various face filters. Brzęczkowski is a Python developer at TrustStamp. Steps to implement face swapping with OpenCV and Python #1 Face detection using histogram of oriented gradients (HOG) Histogram of oriented gradients (HOG) is a feature descriptor that is used to detect objects in computer vision and image processing. Brzęczkowski demonstrated the working of a HOG using square patches which when hovered over an array of images produces a histogram of oriented gradients feature vectors. These feature vectors are then passed to the classifier to generate a result having the highest matching samples. In order to implement face detection using HOG in Python, the image needs to be imported using import OpenCV. Next a frontal face detector object is created for the loaded image detector=dlib.get_frontal_face_detector(). The detector then produces the vector with the detected face. #2 Facial landmark detection aka face alignment Face landmark detection is the process of finding points of interest in an image of a human face. When dlib is used for facial landmark detection, it returns 68 unique fashion landmarks for the whole face. After the first iteration of the algorithm, the value of T equals 0. This value increases linearly such that at the end of the iteration, T gets the value 10. The image evolved at this stage produces the ‘ground truth’, which means that the iteration can stop now. Due to this working, this stage of the process is also called as face alignment. To implement this stage, Brzęczkowski showed how to add a predictor in the Python program with the values shape_predictor_68_face_landmarks.dat such that it produces a model of around 100 megabytes. This process generally takes up a long time as we tend to pick the biggest clearer image for detection. #3 Finding face border using convex hull The convex hull is a set of points defined as the smallest convex polygon, which encloses all of the points in the set. This means that for a given set of points, the convex hull is the subset of these points such that all the given points are inside the subset. To find the face border in an image, we need to change the structure a bit. The structure is first passed to the convex hull function with return points to false, this means that we get an output of indexes. Brzęczkowski then exhibited the face border in the image in blue color using the find_convex_hull.py function. #4 Approximating nonlinear operations with linear operations In a linear filtering of an image, the value of an output pixel is a linear combination of the values of the pixels. Brzęczkowski put forth the example of Affine transformation which is a type of linear mapping method and is used to preserve points, straight lines, and planes. On the other hand, a non-linear filtering produces an output which is not a linear function of its input. He then goes on to unveil both the transitions using his own image. Brzęczkowski then advised users to check the website learnOpenCV.com to learn how to create a nonlinear operation with a linear one. #5 Finding triangles in an image using Delaunay triangulation A Delaunay triangulation subdivides a set of points in a plane into triangles such that the points become vertices of the triangles. This means that this method subdivides the space or the surface into triangles in such a way that if you look at any triangle on the image, it will not have another point inside the triangle. Brzęczkowski then demonstrates how the image developed in the previous stage contained “face points from which you can identify my teeth and then create sub div to the object, insert all these points that I created or all detected.” Next, he deploys Delaunay triangulation to produce a list of two angles. This list is then used to obtain the triangles in the image. Post this step, he uses the delaunay_triangulation.py function to generate these triangles on the images. #6 Blending one face into another To recap, we started from detecting a face using HOG and finding its border using convex hull, followed it by adding mouth points to indicate specific indexes. Next, Delaunay triangulation was implemented to obtain all the triangles on the images. Next, Brzęczkowski begins the blending of images using seamless cloning. A seamless cloning combines the attributes of other cloning methods to create a unique solution to allow “sequence-independent and scarless insertion of one or more fragments of DNA into a plasmid vector.” This cloning method also provides a variety of skin colors to choose from. Brzęczkowski then explains a feature called ‘pass on edit image’ in the Poisson image editing which uses the value of the gradients instead of the identities or the values of the pixels of the image. To implement the same method in OpenCV, he further demonstrates how information like source, destination, source image destination, mask and center (which is the location where the cloned part should be placed) is required to blend the two faces. Brzęczkowski then depicts a  string of illustrations to transform his image with the images of popular artists like Jamie Foxx, Clint Eastwood, and others. #7 Stabilization using optical flow with the Lucas-Kanade method In computer vision, the Lucas-Kanade method is a widely used differential method for optical flow estimation. It assumes that the flow is essentially constant in a local neighborhood of the pixel under consideration, and solves the basic optical flow equations for all the pixels in that neighborhood, by the least-squares criterion. Thus by combining information from several nearby pixels, the Lucas–Kanade method resolves the inherent ambiguity of the optical flow equation. This method is also less sensitive to noises in an image. By using this method to implement the stabilization of the face swapped image, it is assumed that the optical flow is essentially constant in a local neighborhood of the pixel under consideration in human language. This means that “if we have a red point in the center we assume that all the points around, let's say in this example is three on three pixels we assume that all of them have the same optical flow and thanks to that assumption we have nine equations and only two unknowns.” This makes the computation fairly easy to solve. By using this assumption the optical flow works smoothly if we have the previous gray position of the image. This means that for face swapping images using OpenCV, a user needs to have details of the previous points of the image along with the current points of the image. By combining all this information, the actual point becomes a combination of the detected landmark and the predicted landmark. Thus by implementing the Lucas-Kanade method for stabilizing the image, Brzęczkowski implements a non-shaky version of his face-swapped image. Watch Brzęczkowski’s full video to see a step-by-step implementation of a face-swapping task. You can learn advanced applications like facial recognition, target tracking, or augmented reality from our book, ‘Mastering OpenCV 4 with Python’ written by Alberto Fernández Villán. This book will also help you understand the application of artificial intelligence and deep learning techniques using popular Python libraries like TensorFlow and Keras. Getting to know PyMC3, a probabilistic programming framework for Bayesian Analysis in Python How to perform exception handling in Python with ‘try, catch and finally’ Implementing color and shape-based object detection and tracking with OpenCV and CUDA [Tutorial] OpenCV 4.0 releases with experimental Vulcan, G-API module and QR-code detector among others

Bayes' theorem, named after 18th-century British mathematician Thomas Bayes, is a mathematical formula for determining conditional probability. This theorem is used to revise or update existing predictions or theories using new or additional evidence. Bayes theorem is also used in the field of data science as it provides a rule for moving from a prior probability to a posterior probability.  In Bayesian statistics, a prior probability is the probability of an event before a new data is collected and a posterior probability is a conditional probability that is allotted after the relevant evidence is acquired. Hence, the Bayes algorithm is one of the most popular machine learning techniques in the field of data science.  In this post, we are going to discuss a specific Bayesian implementation called probabilistic programming (PP) in Python, considering that modern Bayesian statistics is mainly done by writing code. The probabilistic programming enables flexible specification of complex Bayesian statistical models, thus giving users the ability to focus more on model design, evaluation, and interpretation, and less on mathematical or computational details. Further Reading [box type="shadow" align="" class="" width=""]To know more about Bayesian data analysis techniques using PyMC3 and ArviZ, read our book ‘Bayesian Analysis with Python’, written by Osvaldo Martin. This book will help you acquire skills for a practical and computational approach towards Bayesian statistical modeling. The book also lists the best practices in Bayesian Analysis with the help of sample problems and practice exercises.[/box] A group of researchers have published a paper “Probabilistic Programming in Python using PyMC” exhibiting a primer on the use of PyMC3 for solving general Bayesian statistical inference and prediction problems. PyMC3 is a popular open-source PP framework in Python with an intuitive and powerful syntax closer to the natural syntax statisticians. The PyMC3 installation depends on several third-party Python packages which are automatically installed when installing via pip. It requires four dependencies: Theano, NumPy, SciPy, and Matplotlib. To undertake the full advantage of PyMC3, the researchers suggest, the optional dependencies Pandas and Patsy should also be installed using: pip install patsy pandas. How to use PyMC3 in probabilistic programming? In the paper, the researchers have utilized a simple Bayesian linear regression model with normal priors for the parameters. The unknown variables in the model are also assigned a prior distribution. The artificial data in the model are then simulated using NumPy’s random module, followed by the PyMC3 model to retrieve the corresponding parameters. The straightforward PyMC3 model structure is used to generate the unknown data as it is close to the statistical notation.  Firstly, the necessary components are imported from PyMC to build the required model. It is represented in the full format initially and then explained partly. The paper states, “Following instantiation of the model, the subsequent specification of the model components is performed inside a with statement: with basic_model: This creates a context manager, with our basic model as the context, that includes all statements until the indented block ends.” This means that all the PyMC3 objects introduced in the indented code block below the with statements are added to the model behind the scenes. In the absence of this context manager idiom, users would be forced to manually associate each of the variables with the basic model immediately after we create them. Also, if a user tries to create a new random variable without a with model: statement, it will cause an error due to the absence of an obvious model for the variable to be added to.  Next, to obtain posterior estimates for the unknown variables in the model, the posterior estimates are calculated analytically. The researchers have explained two approaches to obtain posterior estimates, users can choose either of them depending on the structure of the model and the goals of the analysis. The first approach is called finding the maximum a posteriori (MAP) point using optimization methods and the second approach is computing summaries based on samples drawn from the posterior distribution using Markov Chain Monte Carlo (MCMC) sampling methods. For producing a posterior analysis of the required model, PyMC3 provides plotting and summarization functions for inspecting the sampling output.  A simple posterior plot can be created using traceplot. In the traceplot, the left column consists of the smoothed histogram while the right column contains the samples of the Markov chain plotted in sequential order. In addition, the summary function of PyMC3 also provides a text-based output of common posterior statistics. You can also learn more about the practical implementation of PyMC3 and its loss functions in the book ‘Bayesian Analysis with Python’ by Packt Publishing. How Facebook data scientists use Bayesian optimization for tuning their online systems How to perform exception handling in Python with ‘try, catch and finally’ Fake Python libraries removed from PyPi when caught stealing SSH and GPG keys, reports ZDNet Netflix open-sources Metaflow, its Python framework for building and managing data science projects ActiveState adds thousands of curated Python packages to its platform

Deep learning, an emerging branch of machine learning, has garnered a lot of recognition in the field of technology over the last decade. It is regarded as a game-changer in AI, with distinct progress in computer vision, natural language processing (NLP), speech and other areas of machine learning. This year an Indeed survey found ‘deep learning engineer’ to be the best job in a tech position in the USA. Though deep learning has many benefits and a very appealing track record, not everybody can afford deep learning. It has some downsides like large data requirements, being excessively expensive, and has a high computing time. Below is a breakdown of Rachael Tatman’s talk “Put down the deep learning: When not to use neural networks and what to do instead” at the PyCon 2019 conference that delved into the problems with deep learning. Tatman is a data science advocate at Kaggle. Deep learning models require a very large amount of data in order to perform better than other techniques. Also, according to Tatman, just the compute of a simple image generation model in deep learning can cost around $60,000. This cost will increase with the complexity of the data models. It additionally requires expensive GPUs and hundreds of machines which will again deepen the cost to the user. Many less skilled people also find it difficult to adopt deep learning, as there is no standard theory available for learning about deep learning tools. The choice of a deep learning tool depends on the user’s knowledge of topology, training method, and other parameters. Next, deep learning also takes a lot of time for training large models. As the talk progresses, Tatman provides a list of three different types of models that can be used instead of deep learning. The three proposed models are regression-based models, tree-based models, and distance-based models.  The three proposed models instead of deep learning The most interpretable: Regression-based models The biggest advantage of a regression-based model is that it has a “well-principled” understanding of problems and provides many kinds of regression models, unlike deep learning. Users can simply work through the flowchart and decide on the best type of regression model for their data.  Some other advantages of regression models include its “fast to fit” feature. This means that it is much faster to fit when compared to a neural network, especially “if you're working with a well-optimized library the Python regression libraries tend to vary wildly so you might want to do a little bit of shopping around”. It also works well with small data as Tatman affirmed that she has worked on eight dozen data points. She added that since regression models are easy to interpret, she was able to learn many useful and interesting things from the data.  A few drawbacks of regression models are that a bit more data preparation is needed than for some other methods. They also require validation as regression models are based on strong assumptions about the distribution of the data points or the distribution of the errors.  Tatman also proclaimed that if she were to use a single machine learning model for the rest of her life, it would be a mixed-effects regression model. Mixed-effects models are extensions of linear regression models for data that are collected and summarized in groups. It is mainly used to determine the expected or mean values of the subject population. She believes, “you need to do a little bit more hands-on stuff, you need to do your validation, you probably need to do some additional data cleaning,” however, it only takes some time to do a lot of computing in less money and data. Want to know more about Regression? [box type="shadow" align="" class="" width=""]With so many benefits in regression-based models, you should definitely give Regression models a try. Read our book ‘Python Machine Learning By Example’ written by Yuxi (Hayden) Liu, to learn about regression algorithms and their evaluation. You can also master the art of building your own machine learning systems using other models such as Support Vector Machines and Text Analysis Algorithms with this example-based practical guide.[/box] The user-friendliest: Tree-based models  The next model which has the ability to replace deep learning models is called the tree-based models works that similar to a decision tree. It checks each node for a feature and depending on the value of that feature, the user can decide the path to be followed. When going down a particular path, it again checks for nodes with a feature. In this way, it works recursively to cut down a decision region into smaller chunks. Tatman also notified that developers generally opt for a forests model, instead of a tree-based model. A random forest is an ensemble model that combines many different decision trees together into a single model.  Per Tatman, “If you're in the machine learning community you might actually associate random forests with Kaggle and from 2010 to 2016, about two-thirds of all Kaggle competition winners used random forests.” On the other hand, “less than half use some form of deep learning, also random forests continue to do very well today.”  In the case of classification of data, random forests deliver better performance than logistic regression. It also does not need a lot of data cleaning or model validation. Random forests also do not require a user to convert the categorical variables, it simply undertakes the values and provides a corresponding output. It also supports many easy to use packages like XG boost, LightGBM, CatBoost, and others. In short, regression trees are the most user-friendly model, especially when doing classification. The drawbacks of trees/random forests are that they can easily overfit, it is also more sensitive to differences between datasets. It is less interpretable and requires more compute and training time when compared to regression models. Thus, tree-based models require little money but do need some data and time to train big data sets. The most lightweight: Distance-based models The last model, which according to Tatman can replace deep learning models is a common notation to group together a large group of methods like K-nearest neighbors, Gaussian Mixture models, and Support Vector machine. These models work with the basic idea that “points closer together to each other in a particular feature space are more likely to be in the same group.” The K-nearest neighbor model decides the value of a point based on the nearest majority neighbors. The Gaussian mixture models utilizes any distribution of distribution points that are a mixture of different Gaussians. The support vector model tries to be as far away from all the data points as possible. Distance-based models, particularly support vector models work very well with small data sets. They also tend to train 10 times faster than a regression model on the same data. In terms of accuracy, distance-based models lag behind other models, but in case of quick and dirty modeling, they perform better. They are good at data classification but are a little slower when compared to regression-based models. Consequently, distance-based models take very little time, requires very little money and are extremely lightweight. To conclude, Tatman says that the choice of one’s model should depend on the kind of time and money, the individual or organization possesses. Also, the most vital point to choose a model depends on its performance. Tatman adds, “based on empirical evidence right now it looks like deep learning will perform the best on a given data set given sufficient time money and compute.” Watch Tatman’s full talk for a detailed comparison of the three models. You can learn more about all the above machine learning models from our book, ‘Python Machine Learning By Example’ written by Yuxi (Hayden) Liu. The book will help you in implementing machine learning classification and regression algorithms from scratch in Python. Also, learn how to optimize the performance of a machine learning model for your application from our book. François Chollet, creator of Keras on TensorFlow 2.0 and Keras integration, tricky design decisions in Deep Learning, and more Baidu adds Paddle Lite 2.0, new development kits, EasyDL Pro, and other upgrades to its PaddlePaddle deep learning platform Why use JVM (Java Virtual Machine) for deep learning Prof. Rowel Atienza discusses the intuition behind deep learning, advances in GANs & techniques to create cutting-edge AI models Why Intel is betting on BFLOAT16 to be a game changer for deep learning training? Hint: Range trumps Precision.

This blog post will examine a hypothetical dataset of website visits and customer conversion, to illustrate how decision trees are a more flexible mathematical model than linear models such as logistic regression. Imagine you are monitoring the webpage of one of your products. You are keeping track of how many times individual customers visit this page, the total amount of time they've spent on the page across all their visits, and whether or not they bought the product. Your goal is to be able to predict, for future visitors, how likely they are to buy the product, based on the page visit data. You are considering presenting a discount, or some other kind of offer, to customers you think are likely to buy the product but haven't yet. Get to know more about decision trees and linear models! [box type="shadow" align="" class="" width=""]If you are interested in building your knowledge to prepare data for regularized logistic regression and random forest algorithms, read our book Data Science Projects with Python written by Stephen Klosterman. This book will give you practical guidance on industry-standard data analysis and machine learning tools in Python, with the help of realistic data. You will also learn how to use pandas and Matplotlib to critically examine a dataset with summary statistics and graphs and extract the insights you seek to derive. [/box] After logging the data on many customers, you visualize them and see the following, including some jitter to help see all the data points: There are several interesting patterns visible here. We see that in general, the longer someone spends on the page, the more likely they are to purchase the item. However, this effect seems to depend on the number of visits, in a complex way. Someone who visited the page once and spent at least two minutes there (i.e. two minutes per visit) seems likely to buy, at least up until 18 or so minutes. But someone who visited 10 times as much as this seems likely to buy after only 12 minutes cumulative time (1.2 minutes per visit). Additionally, there is a phenomenon of customers who spend a relatively long time (at least 18 or 19 minutes) over a relatively small number of visits (just one or two), who don't buy. Maybe they opened the page, but then walked away from their computer, and closed the page as soon as they came back. Whatever the reason, the patterns in this data set are interesting and complicated. If you want to create a predictive model of these data, you should consider the likely success of non-linear models, such as decision trees, versus linear models, such as logistic regression (for more information see chapter 3 of my book, Data Science Projects with Python). Logistic Regression as a linear model At a high level, linear models will take the feature space (the two-dimensional space where time is on the x-axis and number of visits is on the y-axis, as in the graph above), and seek to draw a straight line somewhere that creates an accurate division of the two classes of the response variable ("Bought" or "Did not buy"). Consider how likely this is to work. Where would you draw a straight line on the graph above, so that the two regions on either side of the line would primarily contain responses of only one class? It should be apparent that this is not likely to be an entirely successful task. The best you could probably do would be to draw a line that isolates non-buying customers who spent relatively little time on the page, represented by the region of dots to the left of the graph, from the blue dots representing buying customers to the right. While this would basically ignore the little group of customers to the lower right, it's probably the best you could do overall for most customers, using the straight-line approach. In fact, this is essentially what a logistic regression classifier looks like when the model is calibrated to these data. The above graph shows the regions of prediction ("Unlikely to buy" and "Likely to buy") as red or blue shading in the background. Deeper colors indicate a higher likelihood for either class. The conceptual straight-line decision boundary that divides the two regions mentioned above, would run right through the white portion of the background, where the probability of belonging to either class is very low. In other words, the model is "uncertain" about what prediction to make in this region. From the above graph, it can be seen that in addition to ignoring the small group of non-buying customers in the lower right, a straight line is also not a great model for isolating the non-buying customers on the left of the graph. While you can imagine that a curve might be able to define this boundary, a straight line cannot. Decision Trees as a non-linear model How can we do better? Enter non-linear models. Decision trees are a prime example of non-linear models. Decision trees work by dividing the data up into regions based on the "if-then" type of questions. For example, if a user spends less than three minutes over two or fewer visits, how likely are they to buy? Graphically, by asking many of these types of questions, a decision tree can divide up the feature space using little segments of vertical and horizontal lines. This approach can create a much more complex decision boundary, as shown below. It should be clear that decision trees can be used with more success, to model this data set. Given this, you would have a better model for the likelihood of customer conversion and could then proceed to design offers to increase conversion (for more information see chapter 5 of my book, Data Science Projects with Python). In conclusion, this post has shown how non-linear models, such as decision trees, can more effectively describe relationships in complex data sets than linear models, such as logistic regression. It should be noted that linear models can be extended to non-linearity by various means including feature engineering. On the other hand, non-linear models may suffer from overfitting, since they are so flexible. Nonetheless, approaches to prevent decision trees from overfitting have been formulated using ensemble models such as random forests and gradient boosted trees, which are among the most successful machine learning techniques in use today. As a final caveat, note this blog post presents a hypothetical, synthetic data set, which can be modeled almost perfectly with decision trees. Real-world data is messier, but the same principles hold. I hope you found this conceptual discussion helpful. For a more detailed explanation of how decision trees and logistic regression work "under the hood" with real-world data, and the python code for a similar hypothetical example to that shown here, check out my book Data Science Projects with Python. Author Bio Stephen Klosterman is a machine learning data scientist and the author of the book Data Science Projects with Python. He enjoys helping to frame problems in a data science context and delivering machine learning solutions that business stakeholders understand and value. His education includes a Ph.D. in biology from Harvard University, where he was an assistant teacher of the data science course. About the Book This book Data Science Projects with Python will help you understand the working and output of machine learning algorithms and gain insight into not only the predictive capabilities of the models but also their reasons for making these predictions. The book also provides detailed insight on how to build a classification model and how to conduct a financial analysis to find the optimal threshold for binary classification. This will help you with financial budgeting and operational strategy for a well-optimized usage model. At the end of this book, you will be able to confidently use various machine learning algorithms to perform detailed data analysis. Netflix open-sources Metaflow, its Python framework for building and managing data science projects What does a data science team look like? Get Ready for Open Data Science Conference 2019 in Europe and California How to learn data science: from data mining to machine learning Dr.Brandon explains Decision Trees to Jon

Generative adversarial networks (GANs) have been at the forefront of research on generative models in the last couple of years. GANs have been used for image generation, image processing, image synthesis from captions, image editing, visual domain adaptation, data generation for visual recognition, and many other applications, often leading to state of the art results. One of the tutorials titled, ‘Generative Adversarial Networks’ conducted at the CVPR 2018 (a Conference on Computer Vision and Pattern Recognition held at Salt Lake City, USA) provides a broad overview of generative adversarial networks and how GANs can be trained to perform different purposes.  The tutorial involved various speakers sharing basic concepts, best practices of the current state-of-the-art GAN including network architectures, objective functions, other training tricks, and much more. Let us look at how GANs are trained for different use cases.  There’s more to GANs….. If you further want to explore different examples of modern GAN implementations, including CycleGAN, simGAN, DCGAN, and 2D image to 3D model generation, you can explore the book, Generative Adversarial Networks Cookbook written by Josh Kalin. The recipes given in this cookbook will help you build on a common architecture in Python, TensorFlow and Keras to explore increasingly difficult GAN architectures in an easy-to-read format. Training GANs for object detection using Adversarial Learning Xialong Wang, from Carnegie Mellon University talked about object detection in computer vision as well as from the context of taking actions in robots. He also explained how to use adversarial learning for instances beyond image generation. To train a GAN, the key idea is to find the adversarial tasks for your target tasks to improve your target by fighting against these adversarial tasks. In computer vision if your target task is to recognize a bird using object detection, one adversarial task is adding occlusions by generating a mask to accrue the bird’s head and its leg which will make it difficult for the detector to recognize. The detector will further try to conquer these difficult tasks and from then on it will become robust to Occlusions. Another adversarial task for object detection can be Deformations. Here the image can be slightly rotated to make the detection difficult.  For training robots to grasp objects, one of the adversaries would be the Shaking test. If the robot arm is stable enough the object it grasps should not fall even with a rigourous shake. Another example is snatching. If another arm can snatch easily, it means it is not completely trained to resist snatching or stealing. Wang said the CMU research team tried generating images using DCGAN on the COCO dataset. However, the images generated could not assist in training the detector as the detectors could easily detect them as false images. Next, the team generated images using Conditional GANs on COCO but these didn’t help either. Hence, the team generated hard positive examples in feed by adding real world occlusions or real world deformations to challenge the detectors. He then talked about a Standard Fast R-CNN Detector which takes an image input in the convolutional neural network language model. After taking the input, the detector extracts features for the whole image, and later you can crop the features according to the proposal bounding box. These cropped features are resized to channel (C*6*6); here 6*6 is interred spatial dimensions. These features are the object features you want to focus on and can also use them to perform classification or regression for detections. The team has added a small network in the middle that would input the extracted features and generate a mask. The mask will assist which spatial locations to chop out certain features that would make it hard for the detectors to recognize. He also shared the benchmark results of the tests using different datasets like the AlexNet, VGG16, FRCN, and so on. The ASTN and the ASDN model showed improved output over the other networks.   Understanding Generative Adversarial Imitation Learning (GAIL) for training a machine to imitate human behaviours Stefano Ermon from Stanford University explained how to use Generative modeling ideas and GAN training to imitate human behaviours in complex environments.  A lot of progress in reinforcement learning has been made with successes in playing board games such as Chess, video games, and so on. However, Reinforcement Learning has one limitation. If you want to use it to solve a new task you have to specify a cost signal / a reward signal to provide some supervision to your reinforcement learning algorithm. You also need to specify what kind of behaviors are desirable and which are not.   In a game scenario the cost signal is whether you win or you lose. However, in further complex tasks like driving an autonomous vehicles to specify a cost signal becomes difficult as there are different objective functions like going off road, not moving above the speed limit, avoiding a road crash, and much more.  The simplest method one can use is Behavioural cloning where you can use your trajectories and your demonstrations to construct a training set of states with the corresponding action that the expert took in those states. You can further use your favorite supervised learning method classification or regression if the actions are continuous. However, this has some limitations: Small errors may compound over time as the learning algorithm will make certain mistakes initially and these mistakes will lead towards never seen before states or objects. It is like a Black box approach where every decision requires initial planning. Ermon suggests an alternative to imitation could be an Inverse RL (IRL) approachHe also demonstrates the similarities between RL and IRL. For the complete demonstration, you can check out the video.  The main difference between a GAIL and GANs is that in GANs the generator is taking inputs, random noise and maps them to the neural network producing some samples for the detector. However, in GAIL, the generator is more complex as it includes two components, a policy P which you can train and an environment (Black Box simulator) that can’t be controlled. What matters is the distribution over states and actions that you encounter when you navigate the environment using the policy that can be tuned. As the environment is difficult to control, training the GAIL model is harder than the simple GANs model. On the other hand, in a GANs model, training the policy is challenging such that the discriminator goes into the direction of fooling.  However, GAIL is the easier generative modelling task because you don’t have to learn the whole thing end to end and neither do you have to come up with a large neural network that maps noise into behaviours as some part of the input is given by the environment. But it is harder to train because you don't really know how the black box works. Ermon further explains how using Generative Adversarial Imitation Learning, one can not only imitate complex behaviors, but also learn interpretable and meaningful representations of complex behavioral data, including visual demonstrations with a method named as InfoGAN, a method, built on top of GAIL.   He also explained a new framework for multi-agent imitation learning for general Markov games by integrating multi-agent RL with a suitable extension of multi-agent inverse RL. This method will generalize Generative Adversarial Imitation Learning (GAIL) in the single agent case. This method will successfully imitate complex behaviors in high-dimensional environments with multiple cooperative or competing agents. To know more about further demonstrations on GAIL, InfoGAIL, and Multi-agent GAIL, watch the complete video on YouTube. Knowing the basics isn’t enough, putting them to practice is necessary. If you want to use GANs practically and experiment with them, Generative Adversarial Networks Cookbook by Josh Kalin is your go-to guide. With this cookbook, you will work with use cases involving DCGAN, Pix2Pix, and so on. To understand these complex applications, you will take different real-world data sets and put them to use. Prof. Rowel Atienza discusses the intuition behind deep learning, advances in GANs & techniques to create cutting edge AI- models Now there is a Deepfake that can animate your face with just your voice and a picture using temporal GANs Now there’s a CycleGAN to visualize the effects of climate change. But is this enough to mobilize action?

TensorFlow 2.0 was released recently with tighter integration with Keras, eager execution enabled by default, three times faster training performance, a cleaned-up API, and more.  TensorFlow 2.0 had a major API Cleanup. Many API symbols are removed or renamed for better consistency and clarity. It now enables eager execution by default which effectively means that your TensorFlow code runs like numpy code. Keras has been introduced as the main high-level API to enable developers to easily leverage Keras’ various model-building APIs. TensorFlow 2.0 also has the SavedModel API that allows you to save your trained Machine learning model into a language-neutral format.  In May, Paige Bailey, Product Manager (TensorFlow) and Laurence Moroney,  Developer Advocate at Google sat down to discuss frequently asked questions on TensorFlow 2.0. They talked about TensorFlow prebuilt binaries, the TF 2.0 upgrade script, Tensorflow Datasets, and Python support. Can I ask about any prebuilt binary for the RTX 2080 GPU on Ubuntu 16?  Prebuilt binaries for TensorFlow tend to be associated with a specific driver from Nvidia. If you're taking a look at any of the prebuilt binaries, take a look at what driver or what version of the driver you have supported on that specific card. It's easy for you to go to the driver vendor and download the latest version. But that may not be the one that TensorFlow is built for or the one that it supports. So, just make sure that they actually match each other.  Do my TensorFlow scripts work with TensorFlow 2.0?  Generally, TensorFlow scripts do not work with TensorFlow 2.0. But TensorFlow 2.0 has created an upgrade utility that is automatically downloaded with TensorFlow 2.0. For more information, you can check out this medium blog post that Paige and her colleague Anna created. It shows how you can upgrade script on an end file - any arbitrary Python file or even Jupyter Notebooks. It'll give you an export.txt file that shows you all of the symbol renames, the added keywords, and then some manual changes.  When will TensorFlow be supported in Python 3.7 and hence be accessed in Anaconda 3? TensorFlow has made the commitment that as of January 1, 2020, they no longer support Python 2. They are firmly committed to Python 3 and Python 3 support.  Is it possible to run Tensorboard on colabs? You can run Tensorboard on colabs and do different operations like smoothing, changing some of the values, and using the embedding visualizer directly from your collab notebook in order to understand accuracies and to be able to model performance debugging. You also don't have to specify ports which means you need not remember to have multiple tensor board instances running. Tensorboard automatically selects one that would be a good candidate.  How would you use [TensorFlow’s] feature_columns with Keras? TensorFlow's feature_columns API is quite useful for non-numerical feature processing. Feature columns are a way of getting your data efficiently into Estimators and you can use them in Keras. TensorFlow 2.0 also has a migration guide if you wanted to migrate your models from using Estimators to being more of a TensorFlow 2.0 format with Keras.   What are some simple data sets for testing and comparing different training methods for artificial neural networks? Are there any in TensorFlow 2.0? Although MNIST and Fashion-MNIST are great, TensorFlow 2.0 also has TensorFlow Datasets which provide a collection of datasets ready to use with TensorFlow. It handles downloading and preparing the data and constructing a tf.data. TensorFlow Datasets is compatible with both TensorFlow Eager mode and Graph mode. Also, you can use them with all of your deep learning and machine learning models with just a few lines of code.  What about all the web developers who are new to AI, how does TensorFlow 2.0 help them get started? With TensorFlow 2.0, the web models that you create using saved model can be deployed to TFLite, or TensorFlow.js. The Keras layers are also supported in TensorFlow.js, so it's not just for Python developers but also for JS developers or even R developers.  You can watch Paige and Lawrence answering more questions in this three-part video series available on YouTube. Some of the other  questions asked were: Is there any TensorFlow.js transfer learning example for object detection? Are you going to publish the updated version of TensorFlow from Poets tutorial from Pete Warden implementing TF2.0. TFLite 2.0 and NN-API for faster inference on Android devices equipped with NPU/DSP? Will the frozen graph generated from TF 1.x work on TF 2.0? Which is the preferred format for saving the model GOIU forward saved_model (SM) or hd5? What is the purpose of keeping Estimators and Keras as separate APIs?  If you want to quickly start with building machine learning projects with TensorFlow 2.0, read our book TensorFlow 2.0 Quick Start Guide by Tony Holdroyd. In this book, you will get acquainted with some new practices introduced in TensorFlow 2.0. You will also learn to train your own models for effective prediction, using high-level Keras API.  TensorFlow.js contributor Kai Sasaki on how TensorFlow.js eases web-based machine learning application development Introducing Spleeter, a Tensorflow based python library that extracts voice and sound from any music track. TensorFlow 2.0 released with tighter Keras integration, eager execution enabled by default, and more! Brad Miro talks TensorFlow 2.0 features and how Google is using it internally

TensorFlow 2.0, released in October, has got developers excited about a myriad of features and its ease of use.  At the EuroPython Conference 2019, Brad Miro, developer programs engineer at Google talked about the updates being made to TensorFlow 2.0. He also gave an overview of how Google is using TensorFlow, moving on to why Python is important for TensorFlow development and how to migrate from TF 1.x to TF 2.0. EuroPython is one of the most popular Python programming language community conferences. Below are some highlights from Brad’s talk at EuroPython. What is TensorFlow? TensorFlow, an open-source deep learning library developed at Google, first released in 2015. It’s a Python framework that includes a number of utilities for helping you write deep neural networks supporting both GPUs and TPUs. A lot of deep learning involves using mathematics, statistics, and algebra and perform low-level optimizations with your system. TensorFlow removes a lot of those abstractions leaving you to focus on actually writing your model. How TensorFlow is used internally at Google Tensorflow is used internally at Google to power all of its machine learning and AI. Google’s data centers are powered using AI and TensorFlow to help optimize the usage of these data centers to reduce bandwidth, to ensure network connections are optimized, and to reduce power consumption. TensorFlow also is useful for performing global localization in Google Maps. It is also used heavily in the Google Pixel range of smartphones to help optimize the software. These technologies are also used in medical research specifically in the field of Computer Vision. For example, Tensorflow is used to distinguish between the retinal image of a healthy eye from the retinal image of an eye that has diabetic retinopathy.   Further Learning If you want to learn to build more computer vision applications with TensorFlow 2.0, check out the book Hands-On Computer Vision with TensorFlow 2 by Benjamin Planche, and Eliot Andres. This book by Packt Publishing is a practical guide to building high-performance systems for object detection, segmentation, video processing, smartphone applications, and more. By the end of the book, you will have both the theoretical understanding and practical skills to solve advanced computer vision problems with TensorFlow 2.0. Furthermore, Google is using AI and TensorFlow to predict whether or not objects in space are planets. To summarize, they use AI to predict whether or not fluctuations in the brightness of an object is due to it being a planet.   Why Python is so important for TensorFlow Python has always been the choice for TensorFlow due to the language being extremely easy to use and having a rich ecosystem for data science including tools such as numpy, scikit-learn, and pandas. When TensorFlow was being built, the idea was that it should have the simplicity of numpy, performance of C but ease of use of Python.  What does TensorFlow 2.0 bring to the table TensorFlow 2.0 is powerful, flexible, scalable and easily deployable.  What’s gone Session.run tf.control_dependencies tf.global_variables_initializer tf.cond, tf.while_loop Tf.contrib What’s new Eager execution enabled by default  tf.function Keras as main high level API Distribution Strategy API SavedModel API  TensorFlow 2.0 had a major API Cleanup. Many API symbols are removed or renamed for better consistency and clarity.  Session.run has been replaced with eager execution which effectively means that your tensorflow code runs like numpy code.  Eager execution enables fast iteration and intuitive debugging without building a graph. It also makes creating and experimenting with models using TensorFlow easier. It can be especially useful when using the tf.keras model subclassing API. TensorFlow 2.0 has tf.function, a python decorator that lets you run regular Python code which is later compiled down to TensorFlow code using AutoGraph. The Distribution Strategy API in TensorFlow 2.0 allows machine learning researchers to distribute training across a wide variety of compute configurations. This release also allows distributed training with Keras’ model.fit and custom training loops. Keras is introduced as the main high-level API. Keras is a popular high-level API used for easy and fast prototyping, building, and training of deep learning models. This will enable developers to easily leverage their various model-building APIs. Using Keras with TensorFlow has two main methods.  Symbolic (Keras sequential) Your model is a graph of layers Any graph you compile will run  TensorFlow helps you debug by catching errors at compile time Imperative method (Keras subclassing) Your model is Python bytecode Complete flexibility and control  Harder to debug/ Harder to maintain  There are pros and cons of using each method; it really just depends on what your specific use cases are. The SavedModel API allows you to save your trained ML model into a language-neutral format. With TensorFlow 2.0, all TensorFlow ecosystem projects including TensorFlow Lite, TensorFlow JS, TensorFlow Serving, and TensorFlow Hub, support SavedModels. On Tensorflow Hub, you can store and download pre-built models. You can use TensorFlow Extended which is a Python library that can be run on your servers to productionalize your models. TensorFlow Lite lets you run your TensorFlow models on edge devices. With TensorFlow.js, you can run machine learning models using javascript in the browser or run them on servers using node. TensorFlow also has Swift for TensorFlow to help developers use Swift to develop machine learning models. “Swift for TensorFlow provides a new programming model that combines the performance of graphs with the flexibility and expressivity of Eager execution, with a strong focus on improved usability at every level of the stack. This is not just a TensorFlow API wrapper written in Swift — we added compiler and language enhancements to Swift to provide a first-class user experience for machine learning developers.”  Other packages that exist in the TensorFlow ecosystem used for niche use cases are TF Probability, TF Agents (reinforcement learning), Tensor2Tensor, TF Ranking, TF Text (natural language processing), TF Federated, TF privacy and more.  How to upgrade from TensorFlow 1.x to TensorFlow 2.0 There are several migration guides available on TensorFlow’s website. You can also use the tf.compat.v1 library for backwards compatibility and the tf_upgrade_v2 script which you can execute on top of any Python script to convert TF 1.x code to 2.0 code. You can also read more about TF 2.0 migration in our book Hands-On Computer Vision with TensorFlow 2 which introduces the automatic migration tool  and compares TensorFlow 1 concepts with their TensorFlow 2 counterparts with a detailed guide on migrating to idiomatic TensorFlow 2 code. You can watch Brad’s full talk on YouTube. This video is licensed under the CC BY-NC-SA 3.0 license.  TensorFlow.js contributor Kai Sasaki on how TensorFlow.js eases web-based machine learning application development Introducing Spleeter, a Tensorflow based python library that extracts voice and sound from any music track. TensorFlow 2.0 released with tighter Keras integration, eager execution enabled by default, and more!

Built by Facebook engineers and researchers, Pytorch is an open-source Python-based deep learning framework for developing new machine learning models, explore neural network architecture and deploy them at scale in production.  PyTorch is known for its advanced indexing and functions, imperative style, integration support, and API simplicity. This is one of the key reasons why developers prefer this framework for research and hackability. PyTorch is also the second-fastest-growing open source project on the GitHub community which includes anybody from developers starting to get acquainted with AI to some of the best known AI researchers and some of the best-known companies doing AI.  At its F8 annual developer conference, Facebook shared how production-ready PyTorch 1.0 is being adopted by the community and the industry. If you want to learn how you can use this framework to build projects in machine intelligence and deep learning, you may go through our book PyTorch Deep Learning Hands-On by authors Sherin Thomas and Sudhanshu Passi. This book demonstrates numerous examples and dynamic AI applications and demonstrates the simplicity and efficiency of PyTorch.  A number of companies are using PyTorch for research and for production. At F8 developer conference this year, Jerome Pesenti, Vice President of AI at Facebook introduced representatives from Microsoft, Airbnb, Genentech, and Toyota Research Institute who talked about how the framework is helping them build, train, and deploy production-ready AI. Below are some excerpts from their talks. Read also: How PyTorch is bridging the gap between research and production at Facebook: PyTorch team at F8 conference How Microsoft uses PyTorch for its language modeling service David Aronchick, Head of Open Source Machine Learning Strategy at Microsoft Azure  At Microsoft, PyTorch is being used in their language modeling service. Language modeling service uses state-of-the-art language models for both 1 P (first-party) and 3 P (third party). Microsoft explored a number of deep learning frameworks but was running into several issues. These included a slow transition from research to production, inconsistent and frequently changing APIs, and a trade-off between high-level ease-of-use and low-level flexibility.  To overcome these issues, in partnership with Facebook Microsoft built an internal language modeling toolkit on top of PyTorch. Using the native extensibility that PyTorch provided, Microsoft was able to build advanced/custom tasks and architecture. It also improved the onboarding of new users and was an active and inviting community. As a result of this work, Microsoft was able to scale the language modeling features to billions of words. It also led to intuitive, static, and consistent APIs which resulted in seamless migration from Language modeling toolkit v0.4 to 1.0. They also saw improvements in model sizes. Microsoft have partnered with ics.ai to deliver conversational AI bots across the public sector in the UK. ICS.ai, based in Basingstoke, have trained their Microsoft AI driven chat bots to scale to the demands of large county councils, healthcare trusts and universities. How Airbnb is using conversational AI tools in PyTorch to enhance customer experience Cindy Chen, Senior machine learning Data Scientist at Airbnb Airbnb has built a dialog assistant to integrate smart replies and enhance their customer experience. The core of their Dialog assistant for customer service at Airbnb is powered by PyTorch. They have built the smart replies recommendation model by treating it as a machine translation problem.  Airbnb is translating the customer's input message into agent responses by building a sequence to sequence model. They leverage PyTorch’s Open neural machine translation library to build the sequence to sequence model.  Using Pytorch has significantly sped up the Airbnb’s model development cycle as PyTorch provides them with state-of-the-art technologies such as various attention mechanisms and beam search.  How Genentech uses Pytorch in drug discovery and cancer therapy Daniel Bozinov, Head of AI - Early clinical development informatics, Genentech At Genentech, PyTorch is being used to develop personalized cancer medicine as well as for drug discovery and in cancer therapy.  For drug development, Genentech has built deep learning models for specific domains to make some predictions about the properties of molecules such as toxicity. They're also applying AI to come up with new cancer therapies. They identify unique molecules specific to cancer cells that are only produced by those cancer cells, potentially sensitizing the immune system to attack those cancer cells and basically treat them like an infection.   PyTorch has been their deep learning framework of choice because of features such as easier debugging, more flexible control structures, being natively pythonic, and it’s Dynamic graphs which yield in faster execution. Their model architecture is inspired by textual entailment in natural language processing. They use a partially recurrent neural network as well as a straightforward feed-forward network, combine the outputs of these two networks and predict the peptide binding. Toyota Research Institute adds new driver support features in cars Adrien Gaidon, Machine Learning Lead, Toyota Research Institute Toyota developed a cutting-edge cloud platform for distributed deep learning on high-resolution sensory inputs, especially video. This was designed to add new driver support features to the cars. PyTorch was instrumental in scaling up Toyota’s deep learning system because of features like simple API, integration with the global Python ecosystem, and overall a great user experience for fast exploration. It’s also fast for training on a very large scale. In addition to amping up TRI’s creativity and expertise, Pytorch has also amplified Toyota’s capabilities to iterate quickly from idea to real-world use cases. The team at TRI is excited for new Pytorch production features that will help them accelerate Toyota even further.  In this post, we have only summarized the talks. At F8, these researchers spoke in length about each of their company’s projects and how PyTorch has been instrumental in their growth. You can watch the full video on YouTube.  If you are inspired to build your PyTorch-based deep learning and machine learning models, we recommend you to go through our book PyTorch Deep Learning Hands-On. Facebook releases PyTorch 1.3 with named tensors, PyTorch Mobile, 8-bit model quantization, and more François Chollet, creator of Keras on TensorFlow 2.0 and Keras integration, tricky design decisions in Deep Learning, and more PyTorch announces the availability of PyTorch Hub for improving machine learning research reproducibility

TensorFlow 2.0 was made available in October. One of the major highlights of this release was the integration of Keras into TensorFlow. Keras is an open-source deep-learning library that is designed to enable fast, user-friendly experimentation with deep neural networks. It serves as an interface to several deep learning libraries, most popular of which is TensorFlow, and it was integrated into TensorFlow main codebase in TensorFlow 2.0. In September, Lex Fridman, Research scientist at MIT popularly known for his podcasts, spoke to François Chollet, who is the author of Keras on Keras, Deep Learning, and the Progress of AI. In this post, we have tried to highlight François’ views on the Keras and TensorFlow 2.0 integration, early days of Keras and the importance of design decisions for building deep learning models. We recommend the full podcast that’s available on Fridman’s YouTube channel. Want to build Neural Networks? [box type="shadow" align="" class="" width=""]If you want to build multiple neural network architectures such as CNN, RNN, LSTM in Keras, we recommend you to read Neural Networks with Keras Cookbook by V Kishore Ayyadevara. This book features over 70 recipes such as object detection and classification, building self-driving car applications, understanding data encoding for image, text and recommender systems and more. [/box] Early days of Keras and how it was integrated into TensorFlow I started working on Keras in 2015, says Chollet. At that time Caffe was the popular deep learning library, based on C++ and was popular for building Computer Vision projects. Chollet was interested in Recurrent Neural Networks (RNNs) which was a niche topic at that time. Back then, there was no good solution or reusable open-source implementation of RNNs and LSTMs, so he decided to build his own and that’s how Keras started. “It was going to be mostly around RNNs and LSTMs and the models would be defined by Python code, which was going against mainstream,” he adds. Later, he joined Google’s research team working on image classification. At that time, he was exposed to the early internal version of Tensorflow - which was an improved version of Theano. When Tensorflow was released in 2015, he refactored Keras to run on TensorFlow. Basically he was abstracting away all the backend functionality into one module so that the same codebase could run on top of multiple backends. A year later, the TensorFlow team requested him to integrate the Keras API into TensorFlow more tightly.  They build a temporary TensorFlow-only version of Keras that was in tf.contrib for a while. Then they finally moved to TensorFlow Core in 2017. TensorFlow 2.0 gives both usability and flexibility to Keras Keras has been a very easy-to-use high-level interface to do deep learning. However, it lacked in flexibility - Keras framework was not the optimal way to do things compared to just writing everything from scratch. TensorFlow 2.0 offers both usability and flexibility to Keras. You have the usability of the high-level interface but you have the flexibility of the lower-level interface. You have this spectrum of workflows where you can get more or less usability and flexibility,  the trade-offs depending on your needs. It's very flexible, easy to debug, and powerful but also integrates seamlessly with higher-level features up to classic Keras workflows. “You have the same framework offering the same set of APIs that enable a spectrum of workflows that are more or less high level and are suitable for you know profiles ranging from researchers to data scientists and everything in between,” says Chollet. Design decisions are especially important while integrating Keras with Tensorflow “Making design decisions is as important as writing code”, claims Chollet. A lot of thought and care is taken in coming up with these decisions, taking into account the diverse user base of TensorFlow - small-scale production users, large-scale production users, startups, and researchers. Chollet says, “A lot of the time I spend on Google is actually discussing design. This includes writing design Docs, participating in design review meetings, etc.” Making a design decision is about satisfying a set of constraints but also trying to do so in the simplest way possible because this is what can be maintained and expanded in the future. You want to design APIs that are modular and hierarchical so that they have an API surface that is as small as possible. You want this modular hierarchical architecture to reflect the way that domain experts think about the problem. On the future of Keras and TensorFlow. What’s going to happen in TensorFlow 3.0? Chollet says that he’s really excited about developing even higher-level APIs with Keras. He’s also excited about hyperparameter tuning by automated machine learning. He adds, “The future is not just, you know, defining a model, it's more like an automatic model.” Limits of deep learning wrt function approximators that try to generalize from data Chollet emphasizes that “Neural Networks don't generalize well, humans do.” Deep Learning models are like huge parametric and differentiable models that go from an input space to an output space, trained with gradient descent. They are learning a continuous geometric morphing from an input vector space to an output space. As this is done point by point; a deep neural network can only make sense of points in space that are very close to things that it has already seen in string data. At best it can do the interpolation across points. However, that means in order to train your network you need a dense sampling of the input, almost a point-by-point sampling which can be very expensive if you're dealing with complex real-world problems like autonomous driving or robotics.  In contrast to this, you can look at very simple rules algorithms. If you have a symbolic rule it can actually apply to a very large set of inputs because it is abstract, it is not obtained by doing a point by point mapping. Deep learning is really like point by point geometric morphings. Meanwhile, abstract rules can generalize much better. I think the future is which can combine the two. Chollet also talks about self-improving Artificial General Intelligence, concerns about short-term and long-term threats in AI, Program synthesis, Good test for intelligence and more. The full podcast is available on Lex’s YouTube channel. If you want to implement neural network architectures in Keras for varied real-world applications, you may go through our book Neural Networks with Keras Cookbook. TensorFlow.js contributor Kai Sasaki on how TensorFlow.js eases web-based machine learning application development 10 key announcements from Microsoft Ignite 2019 you should know about What does a data science team look like?

An integral part of using Python involves the art of handling exceptions. There are primarily two types of exceptions; Built-in exceptions and User-Defined Exceptions. In such cases, the error handling resolution is to save the state of execution in the moment of error which interrupts the normal program flow to execute a special function or a code which is called Exception Handler. There are many types of errors like ‘division by zero’, ‘file open error’, etc. where an error handler needs to fix the issue. This allows the program to continue based on prior data saved. Source: Eyehunts Tutorial Just like Java, exceptions handling in Python is no different. It is a code embedded in a try block to run exceptions. Compare that to Java where catch clauses are used to catch the Exceptions. The same sort of Catch clause is used in Python that begins with except. Also, custom-made exception is possible in Python by using the raise statement where it forces a specified exception to take place. Reason to use exceptions Errors are always expected while writing a program in Python which requires a backup mechanism. Such a mechanism is set to handle any encountered errors and not doing so may crash the program completely. The reason to equip python program with the exception mechanism is to set and define a backup plan just in case any possible error situation erupts while executing it. Catch exceptions in Python Try statement is used for handling the exception in Python. A Try clause will consist of a raised exception associated with a particular, critical operation. For handling the exception the code is written within the Except Clause. The choice of performing a type of operation depends on the programmer once catching the exception is done. The below-defined program loops until the user enters an integer value having a valid reciprocal. A part of code that triggers an exception is contained inside the Try block. In case of absence of any exceptions then the normal flow of execution continues skipping the except block. And in case of exceptions raising the except block is caught. Checkout the example: The Output will be: Naming the exception is possible by using the ex_info() function that is present inside the sys module. It asks the user to make another attempt for naming it. Any unexpected values like 'a' or '1.3' will trigger the ValueError. Also, the return value of '0' leads to ZeroDivisionError. Exception handling in Python: try, except and finally There are instances where the suspicious code may raise exceptions which are placed inside such try statement block. Again, there is a code that is dedicated to handling such raised exceptions and the same is placed within the Except block. Below is an example of above-explained try and except statement when used in Python. try:   ** Operational/Suspicious Code except for SomeException:   ** Code to handle the exception How do they work in Python: The primarily used try block statements are triggered for checking whether or not there is any exception occurring within the code. In the event of non-occurrence of exception, the except block (Containing the exceptions handling statements) is executed post executing the try block. When the exception matches the predefined name as mentioned in 'SomeException' for handling the except block, it does the handling and enables the program to continue. In case of absence of any corresponding handlers that deals with the ones to be found in the except block then the activity of program execution is halted along with the error defining it. Defining Except without the exception To define the Except Clause isn’t always a viable option regardless of which programming language is used. As equipping the execution with the try-except clause is capable of handling all the possible types of exceptions. It will keep users ignorant about whether the exception was even raised in the first place. It is also a good idea to use the except statement without the exceptions field, for example some of the statements are defined below: try:    You do your operations here;    ...................... except:    If there is an exception, then execute this block.    ...................... else:    If there is no exception then execute this block.  OR, follow the below-defined syntax: try:   #do your operations except:   #If there is an exception raised, execute these statements else:   #If there is no exception, execute these statements Here is an example if the intent is to catch an exception within the file. This is useful when the intention is to read the file but it does not exist. try:   fp = open('example.txt', r) except:   print ('File is not found')   fp.close This example deals with opening the 'example.txt'. In such cases, when the called upon file is not found or does not exist then the code executes the except block giving the error read like 'File is not found'. Defining except clause for multiple exceptions It is possible to deal with multiple exceptions in a single block using the try statement. It allows doing so by enabling programmers to specify the different exception handlers. Also, it is recommended to define a particular exception within the code as a part of good programming practice. The better way out in such cases is to define the multiple exceptions using the same, above-mentioned except clause. And it all boils down to the process of execution wherein if the interpreter gets hold of a matching exception, then the code written under the except code will be executed. One way to do is by defining a tuple that can deal with the predefined multiple exceptions within the except clause. The below example shows the way to define such exceptions: try:    # do something  except (Exception1, Exception2, ..., ExceptionN):    # handle multiple exceptions    pass except:    # handle all other exceptions You can also use the same except statement to handle multiple exceptions as follows − try:    You do your operations here;    ...................... except(Exception1[, Exception2[,...ExceptionN]]]):    If there is an exception from the given exception list,     then execute this block.    ...................... else:    If there is no exception then execute this block.  Exception handling in Python using the try-finally clause Apart from implementing the try and except blocks within one, it is also a good idea to put together try and finally blocks. Here, the final block will carry all the necessary statements required to be executed regardless of the exception being raised in the try block. One benefit of using this method is that it helps in releasing external resources and clearing up the cache memories beefing up the program. Here is the pseudo-code for try..finally clause. try:    # perform operations finally:    #These statements must be executed Defining exceptions in try... finally block The example given below executes an event that shuts the file once all the operations are completed. try:    fp = open("example.txt",'r')    #file operations finally:    fp.close() Again, using the try statement in Python, it is wise to consider that it also comes with an optional clause – finally. Under any given circumstances, this code is executed which is usually put to use for releasing the additional external resource. It is not new for the developers to be connected to a remote data centre using a network. Also, there are chances of developers working with a file loaded with Graphic User Interface. Such situations will push the developers to clean up the used resources. Even if the resources used, yield successful results, such post-execution steps are always considered as a good practice. Actions like shutting down the GUI, closing a file or even disconnecting from a connected network written down in the finally block assures the execution of the code. The finally block is something that defines what must be executed regardless of raised exceptions. Below is the syntax used for such purpose: The file operations example below illustrates this very well: try: f = open("test.txt",encoding = 'utf-8') # perform file operations finally: f.close() Or In simpler terms: try:    You do your operations here;    ......................    Due to any exception, this may be skipped. finally:    This would always be executed.    ...................... Constructing such a block is a better way to ensure the file is closed even if the exception has taken place. Make a note that it is not possible to use the else clause along with the above-defined finally clause. Understanding user-defined exceptions Python users can create exceptions and it is done by deriving classes out of the built-in exceptions that come as standard exceptions. There are instances where displaying any specific information to users is crucial, especially upon catching the exception. In such cases, it is best to create a class that is subclassed from the RuntimeError. For that matter, the try block will raise a user-defined exception. The same is caught in the except block. Creating an instance of the class Networkerror will need the user to use variable e. Below is the syntax: class Networkerror(RuntimeError):    def __init__(self, arg):       self.args = arg   Once the class is defined, raising the exception is possible by following the below-mentioned syntax. try:    raise Networkerror("Bad hostname") except Networkerror,e:    print e.args Key points to remember Note that an exception is an error that occurs while executing the program indicating such events (error) occur though less frequently. As mentioned in the examples above, the most common exceptions are ‘divisible by 0’, ‘attempt to access non-existent file’ and ‘adding two non-compatible types’. Ensure putting up a try statement with a code where you are not sure whether or not the exception will occur. Specify an else block alongside try-except statement which will trigger when there is no exception raised in a try block. Author bio Shahid Mansuri Co-founder Peerbits, one of the leading software development company, USA, founded in 2011 which provides Python development services. Under his leadership, Peerbits used Python on a project to embed reports & researches on a platform that helped every user to access the dashboard that was freely available and also to access the dashboard that was exclusively available. His visionary leadership and flamboyant management style have yield fruitful results for the company. He believes in sharing his strong knowledge base with a learned concentration on entrepreneurship and business. Introducing Spleeter, a Tensorflow based python library that extracts voice and sound from any music track Fake Python libraries removed from PyPi when caught stealing SSH and GPG keys, reports ZDNet There’s more to learning programming than just writing code

PyTorch, the machine learning library which was originally developed as a research framework by a Facebook intern in 2017, has now grown into a popular deep learning workflow. One of the most loved products by Facebook, PyTorch is free, open source, and used for applications like computer vision and natural language processing (NLP).  At the F8 conference held this year, the PyTorch team consisting of Joe Spisak, the project manager for PyTorch at Facebook AI and Dmytro Dzhulgakov, the tech lead at Facebook AI gave a talk on how Facebook is developing and scaling AI experiences with PyTorch.  Spisak describes PyTorch as an eager and graph-based execution that is defined by ‘run’. This means that when a user executes a Python code, it generates a graph on the fly. It is dynamic in nature and allows the compilation of the static graph. The dynamic neural networks are accessible, thus, allowing the user to change the parameters very quickly. This feature comes in handy for applications like control flow in NLP. Another important feature of PyTorch, according to Spisak, is the ability to generate accurately distributed training models that possess close to billion parameters, including the cutting-edge ones. It also has a simple and easy API that is very intuitive by nature. This is one of the qualities of PyTorch which has endeared many developers, claims Spisak.  Become a pro at Deep Learning with PyTorch! If you want to become an expert in building and training neural network models with high speed and flexibility in text, vision, and advanced analytics using PyTorch 1.x, read our book Deep Learning with PyTorch 1.x - Second Edition written by Sri. Yogesh K., Laura Mitchell, et al.  It will give you an insight into solving real-world problems using CNNs, RNNs, and LSTMs, along with discovering state-of-the-art modern deep learning architectures, such as ResNet, DenseNet, and Inception. How PyTorch is bridging the gap between research and production at Facebook Dzhulgakov points out how general advances in AI are driven by innovative research in the fields of academia or industry and why it’s necessary to bridge this big lag between research and production. He says, “If you have a new idea and you want to take it all the way through to deployment, you usually need to go through multiple steps - figure out what the approach is and then find the training data maybe prepare massage it a little bit. Actually, build and train your model after that and then there is this painful step of transferring your model to a production environment which often historically involved reimplementation of a lot of code so you can actually take and deploy it and scale-up.”  According to Dzhulgakov, PyTorch is trying to minimize this big gap by encouraging advances and experimentations in the field, so that the research is brought into production in a few days, instead of months. Challenges in bringing research to production Following are the various classes of challenges associated with bringing research to production, according to the PyTorch team. Hardware efficiency: In case of a tight latency constraint environment, users are required to fit all the hardware into the performance budget. On the other hand, an underused hardware environment can lead to an increase in cost. Scalability: In Facebook’s recent work, Dzhulgakov says, they have trained on billions of public images, thus indicating significant accuracy gains as compared to regular datasets like imageNet. Similarly, when models are taken to inference, it means that billions of inferences per second are running with multiple diverse models sharing the same hardware. Cross-platform: Neural networks are mostly not isolated as they need to be deployed inside their target application. It has a lot of interdependence with the surrounding code and application, thus posing different constraints like the user will not be able to run Python code or the user will have to work on very constrained computer capabilities if running a mobile device, and more. Reliability: A lot of PyTorch jobs run for multiple weeks on hundreds of GPUs, hence it is important to design a reliable software which can tolerate hardware failures and deliver results.  How PyTorch is tackling these challenges In order to tackle the above-listed challenges, Dzhulgakov says Facebook develops systems that can take up a training job and perform optimizations focused on performance for the performance-critical pieces. The system also applies “recipes for reliability” so that the developer written modeling code is automatically transformed. The Jit package comes into the picture here and acts like a key factor that is built to capture the structure of the Python program with minimal changes. The main goal of the Jit package is to make this process almost seamless. He asserts that PyTorch has been successful since it feels like regular programming in Python and most of its users start developing in traditional PyTorch mode (eager mode) just by writing and prototyping in the program. “For the subset of promising models which shall show what results you need to bring to production either scale up, so you can apply techniques provided by Jit to exist mental codes and annotated in order to run in so-called script code.”   The Jit is like a subset of Python with a thread list of request semantics, which allows the user to apply transparent transformations for the eager mode to the user. The annotations include adding a few lines of Python code on top of the function in such a way that it can be done incrementally on function by function or module by module fashion. This hybrid fashion ensures that the model works along the way. Such powerful PyTorch tools permit the user to share the same code base between research and production environments.  Next, Dzhulgakov deduces that the common factor between research and production is that both teams of developers work on the same code base built on top of PyTorch. Thus, they share the codes among the teams that have a common domain like text classification or object detection or reinforcement learning. These developers prototype models, train new algorithms and address new tasks for quickly transitioning this functionality to the opposite environment. Watch the full talk to see Dzhulgakov’s examples of PyTorch bridging the gap between research and production at Facebook. If you want to become an expert at implementing deep learning applications in PyTorch, check out our latest book Deep Learning with PyTorch 1.x - Second Edition written by Sri. Yogesh K., Laura Mitchell, and Et al. This book will show you how to apply neural networks to domains such as computer vision and NLP. It will also guide you to build, train, and scale a model with PyTorch and cover complex neural networks such as GANs and autoencoders for producing text and images. NVIDIA releases Kaolin, a PyTorch library to accelerate research in 3D computer vision and AI Introducing ESPRESSO, an open-source, PyTorch based, end-to-end neural automatic speech recognition (ASR) toolkit for distributed training across GPUs Facebook releases PyTorch 1.3 with named tensors, PyTorch Mobile, 8-bit model quantization, and more Transformers 2.0: NLP library with deep interoperability between TensorFlow 2.0 and PyTorch, and 32+ pretrained models in 100+ languages PyTorch announces the availability of PyTorch Hub for improving machine learning research reproducibility

At day one of the ongoing Amazon re:Invent 2019, there was a flurry of announcements made for AWS. Most importantly, AWS announced the preview launch of Braket, its own quantum computing service following the likes of IBM, Microsoft, and Google. Amazon also released Amazon SageMaker Operators for Kubernetes to help data scientists using Kubernetes to train, tune, and deploy machine learning models in Amazon SageMaker. re:Invent is Amazon’s flagship conference hosted by Amazon Web Services for the global cloud computing community. This year re: Invent is taking place in Las Vegas, December 2-6, 2019. re:Invent 2019 Day One announcements Braket: AWS’ new quantum service in preview now Amazon Braket (named after the common notation for quantum states) is a fully managed service that helps you get started with quantum computing. Braket consists of a full development environment that helps data scientists to: design quantum algorithms from scratch or choose from a set of pre-built algorithms, test these algorithms on simulated quantum computers (including gate based and quantum annealing superconductors, and ion trap hardware) run them on your choice of different quantum hardware technologies ( including D-Wave, IonQ, and Rigetti) Once your tests are complete, you will be automatically notified and your results will be stored in Amazon S3. Amazon Braket publishes event logs and performance metrics such as completion status and execution time to Amazon CloudWatch. To make it easier to develop hybrid algorithms that combine classical and quantum tasks, Amazon Braket helps manage classical compute resources and establish low-latency connections to the quantum hardware. At re:Invent 2019, AWS also launched the Amazon Quantum Solutions Lab, a collaborative research program that connects you with quantum computing experts from Amazon and its technology and consulting partners. They can help you identify potential uses of quantum computing, build internal expertise, and collaborate on programs to design and test quantum algorithms. Braket is available in preview now. Amazon SageMaker Operators for Kubernetes Now developers and data scientists can use Kubernetes to train, tune, and deploy machine learning models in Amazon SageMaker, with the new Amazon SageMaker Operators for Kubernetes. Customers can install these Amazon SageMaker Operators on their Kubernetes cluster to create Amazon SageMaker jobs natively using the Kubernetes API and command-line Kubernetes tools such as ‘kubectl’. Operators can be used to train machine learning models, optimize hyperparameters for a given model, run batch transform jobs over existing models, and set up inference endpoints. With these operators, users can manage their jobs in Amazon SageMaker from their Kubernetes cluster in Amazon Elastic Kubernetes Service EKS. Amazon SageMaker Operators for Kubernetes are available in select AWS regions. AWS DeepComposer, a creative way to learn Machine Learning Amazon has launched AWS DeepComposer, the world’s first machine learning-enabled musical keyboard at re:Invent 2019. AWS DeepComposer is an educational tool to teach people Machine Learning. AWS DeepComposer gives developers of all skill levels a creative way to experience machine learning – music. https://youtu.be/XH2EbK9dQlg You can input a melody by connecting the AWS DeepComposer keyboard to your computer, or play the virtual keyboard in the AWS DeepComposer console. You can generate an original music composition using the pre-trained genre models in the console. You can then publish your tracks to SoundCloud. It is designed specifically to educate developers by means of tutorials, sample code, and training data. These can be used to get started with building generative AI models, all without having to write a single line of code. With AWS DeepComposer, you can train and optimize GAN models to create original music. GAN models pit two different neural networks against each other to produce new and original digital works based on sample inputs. AWS DeepComposer is available in preview now. Amazon Transcribe now extended to healthcare patients Amazon’s automatic speech recognition service Amazon Transcribe is now available for medical speech as announced in re:Invent 2019. Amazon Transcribe Medical allows physicians to easily and quickly dictate their clinical notes and see their speech converted to accurate text in real-time, without any human intervention. Clinicians can use natural speech and do not have to explicitly call out punctuation like “comma” or “full stop”. This text can then be automatically fed to downstream applications such as EHR systems, or to AWS language services such as Amazon Comprehend Medical for entity extraction. To make it work, you need to capture audio using your device’s microphone and send PCM (Pulse-code modulation) audio to a streaming API based on the popular Websocket protocol. This API will respond with a series of JSON blobs with the transcribed text, as well as word-level time stamps, punctuation, etc. Optionally, you can save this data to an Amazon Simple Storage Service (S3) bucket. Amazon Transcribe Medical is available in US East (N. Virginia) and US West (Oregon) regions. Updates to Microsoft Windows Server AWS has released a bring-your-own-license (BYOL) experience for customers as an easier way to bring, and manage, their existing licenses for Microsoft Windows Server and SQL Server to AWS. The new BYOL experience enables customers who want to use their existing Windows Server or SQL Server licenses to seamlessly create virtual machines in EC2, while AWS takes care of managing their licenses to help ensure compliance to licensing rules specified by the customer. Amazon is also providing End-of-Support Migration Program (EMP) for Windows Server. On January 14, 2020, support for Windows Server 2008 and 2008 R2 will end. Having an application that can run only on an unsupported version of Windows Server is problematic as you will no longer get free security patch updates, leaving you vulnerable to security and compliance risks. This new program combines technology with expert guidance, to migrate your legacy applications running on outdated versions of Windows Server to newer, supported versions on AWS. Other updates announced at Amazon re:Invent 2019 Amazon EventBridge Schema Registry is now in preview.  The schema registry stores the structure (schema) of Amazon EventBridge events and maps them to Java, Python, and Typescript bindings so that you can use the events as typed objects. The existing AWS IoT SiteWise preview adds new features such as creating a virtual representation of your facility, monitor production performance metrics and use AWS IoT SiteWise Monitor to visualize the data in real-time. AWS IoT SiteWise Monitor is a new SaaS application that lets you monitor and interact with the data collected and organized by AWS IoT SiteWise. The upcoming AWS DeepRacer Evo car will include a stereo camera and a Light Detection and Ranging (LIDAR) sensor.  The DeepRacer League in 2020 will have 8 additional races in 5 countries. The preview of EC2 Image Builder, a service that makes it easier and faster to build and maintain secure OS images for Windows Server and Amazon Linux 2, using automated build pipelines. Amazon re:Invent will continue throughout this week (the last day is the 6th of December). You can access the Livestream here. Keep checking this space for news on other updates and launches. Amazon EKS Windows Container Support is now generally available Amazon’s hardware event 2019 highlights: a high-end Echo Studio, the new Echo Show 8, and more 10 key announcements from Microsoft Ignite 2019 you should know about

Running Machine Learning applications on the web browser is one of the hottest trends in software development right now. Many notable machine learning projects are being built with Tensorflow.js. It is one of the most popular frameworks for building performant machine learning applications that run smoothly in a web browser. Recently, we spoke with Kai Sasaki, who is one of the initial contributors to TensorFlow.js. He talked about current and future versions of TF.js, how it compares to other browser-based ML tools and his contributions to the community. He also shared his views on why he thinks Javascript good for Machine Learning. If you are a web developer with working knowledge of Javascript who wants to learn how to integrate machine learning techniques with web-based applications, we recommend you to read the book, Hands-on Machine Learning with TensorFlow.js. This hands-on course covers important aspects of machine learning with TensorFlow.js using practical examples. Throughout the course, you'll learn how different algorithms work and follow step-by-step instructions to implement them through various examples. On how TensorFlow.js has improved web-based machine learning How do you think Machine Learning for the Web has evolved in the last 2-3 years? What are some current applications of web-based machine learning and TensorFlow.js? What can we expect in future releases? Machine Learning on the web platform is a field attracting more developers and machine learning practitioners. There are two reasons. First, the web platform is universally available. The web browser mostly provides us a way to access the underlying resource transparently. The second reason is security.raining a model on the client-side means you can keep sensitive data inside the client environment as the entire training process is completed on the client-side itself. The data is not sent to the cloud, making it more secure and less susceptible to vulnerabilities or hacking. In future releases as well, TensorFlow.js is expected to provide more secure and accessible functionalities. You can find various kinds of TensorFlow.js based applications here. How does TensorFlow.js compare with other web and browser-based machine learning tools? Does it make web-based machine learning application development easier? The most significant advantage of TensorFlow.js is the full compatibility of the TensorFlow ecosystem. Not only can a TensorFlow model be seamlessly used in TensorFlow.js, tools for visualization and model deployment in the TensorFlow ecosystem can also be used in TensorFlow.js. TensorFlow 2 was released in October. What are some new changes made specific to TensorFlow.js as a part of TF 2.0 that machine learning developers will find useful? What are your first impressions of this new release? Although there is nothing special related to TensorFlow 2.0, the full support of new backends is actively developed, such as WASM and WebGPU. These hardware acceleration mechanisms provided by the web platform can enhance performance for any TensorFlow.js application. It surely makes the potential of TensorFlow.js stronger and possible use cases broader. On Kai’s experience working on his book, Hands-on Machine Learning with TensorFlow.js Tell us the motivation behind writing your book Hands-on Machine Learning with TensorFlow.js. What are some of your favorite chapters/projects from the book? TensorFlow.js does not have much history because only three years have passed since its initial publication. Due to the lack of resources to learn TensorFlow.js usage, I was motivated to write a book illustrating how to using TensorFlow.js practically. I think chapters 4 - 9 of my book Hands-On Machine Learning with TensorFlow.js provide readers good material to practice how to write the ML application with TensorFlow.js. Why Javascript for Machine Learning Why do you think Javascript is good for Machine Learning? What are some of the good machine learning packages available in Javascript? How does it compare to other languages like Python, R, Matlab, especially in terms of performance? JavaScript is a primary programming language in the web platform so it can work as a bridge between the web and machine learning applications. We have several other libraries working similarly. For example, machinelearn.js is a general machine learning framework running with JavaScript. Although JavaScript is not a highly performant language, its universal availability in the web platform is attractive to developers as they can build their machine learning applications that are “write once, run anywhere”. We can compare the performance by running state-of-the-art machine learning models such as MobileNet or ResNet practically. On his contribution towards TF.js You are a contributor for TensorFlow.js and were awarded by the Google Open Source Peer Bonus Program. What were your main contributions? How was your experience working for TF.js? One of the significant contributions I have made was fast Fourier transformation operations. I have created the initial implementation of fft, ifft, rfft and irfft. I also added stft (short term Fourier transformation). These operators are mainly used for performing signal analysis for audio applications. I have done several bug fixes and test enhancements in TensorFlow.js too. What are the biggest challenges today in the field of Machine Learning and AI in web development? What do you see as some of the greatest technology disruptors in the next 5 years? While many developers are writing Python programming languages in the machine learning field, not many web developers have familiarity and knowledge of machine learning in spite of the substantial advantage of the integration between machine learning and web platform. I believe machine learning technologies will be democratized among web developers so that a vast amount of creativity is flourished in the next five years. By cooperating with these enthusiastic developers in the community, I believe the machine learning on the client-side or edge device will be one of the major contributions in the machine learning field. About the author Kai Sasaki works as a software engineer in Treasure Data to build large-scale distributed systems. He is one of the initial contributors to TensorFlow.js and contributes to developing operators for newer machine learning models. He has also received the Google Open Source Peer Bonus in 2018. You can find him on Twitter, Linkedin, and GitHub. About the book Hands-On Machine Learning with TensorFlow.js is a comprehensive guide that will help you easily get started with machine learning algorithms and techniques using TensorFlow.js. Throughout the course, you'll learn how different algorithms work and follow step-by-step instructions to implement them through various examples. By the end of this book, you will be able to create and optimize your own web-based machine learning applications using practical examples. Baidu adds Paddle Lite 2.0, new development kits, EasyDL Pro, and other upgrades to its PaddlePaddle platform. Introducing Spleeter, a Tensorflow based python library that extracts voice and sound from any music track TensorFlow 2.0 released with tighter Keras integration, eager execution enabled by default, and more!