A machine learning method called "deep learning," which has been widely used in face recognition and other image- and speech-recognition applications, has shown promise in helping astronomers analyze images of galaxies and understand how they form and evolve. In a new study, accepted for publication in Astrophysical Journal and available online, researchers used computer simulations of galaxy formation to train a deep learning algorithm, which then proved surprisingly good at analyzing images of galaxies from the Hubble Space Telescope. The researchers used output from the simulations to generate mock images of simulated galaxies as they would look in observations by the Hubble Space Telescope. The mock images were used to train the deep learning system to recognize three key phases of galaxy evolution previously identified in the simulations. The researchers then gave the system a large set of actual Hubble images to classify.

An increasing need of running Convolutional Neural Network (CNN) models on mobile devices with limited computing power and memory resource encourages studies on efficient model design. A number of efficient architectures have been proposed in recent years, for example, MobileNet, ShuffleNet, and NASNet-A. However, all these models are heavily dependent on depthwise separable convolution which lacks efficient implementation in most deep learning frameworks. In this study, we propose an efficient architecture named PeleeNet, which is built with conventional convolution instead. Meanwhile, PeleeNet is only 66% of the model size of MobileNet.

Developments in Artificial Intelligence (A.I.) are happening faster today than ever before. However, the nature of progress in A.I. is such that massive technological breakthroughs might go unnoticed while smaller improvements get a lot of media attention. Take the case of face recognition technology. The ability of A.I. to recognize faces might seem like a very big deal, but isn't that groundbreaking when you consider the nature of applied A.I. On the other hand, suppose an A.I. is asked to choose between a genre of music, such as R&B or rock. While it may seem like a simple choice, the mathematical algorithm that must be solved before the A.I makes a decision could take hours and days. Most people get their idea of A.I. from Hollywood movies and science fiction.

Using information theoretic concepts to understand and explore the inner organization of deep neural networks (DNNs) remains a big challenge. Recently, the concept of an information plane began to shed light on the analysis of multilayer perceptrons (MLPs). We provided an in-depth insight into stacked autoencoders (SAEs) using a novel matrix-based Rényi's α-entropy functional, enabling for the first time the analysis of the dynamics of learning using information flow in real-world scenario involving complex network architecture and large data. Despite the great potential of these past works, there are several open questions when it comes to applying information theoretic concepts to understand convolutional neural networks (CNNs). These include for instance the accurate estimation of information quantities among multiple variables, and the many different training methodologies.

There are a lot of neat things going on in deep reinforcement learning. One of the coolest things from last year was OpenAI and DeepMind's work on training an agent using feedback from a human rather than a classical reward signal. There's a great blog post about it at Learning from Human Preferences, and the original paper is at Deep Reinforcement Learning from Human Preferences. I've seen a few recommendations that reproducing papers is a good way of levelling up machine learning skills, and I decided this could be an interesting one to try with. It was indeed a super fun project, and I'm happy to have tackled it - but looking back, I realise it wasn't exactly the experience I thought it would be. If you're thinking about reproducing papers too, here are some notes on what surprised me about working with deep RL.

In this post, I'll discuss commonly used architectures for convolutional networks. As you'll see, almost all CNN architectures follow the same general design principles of successively applying convolutional layers to the input, periodically downsampling the spatial dimensions while increasing the number of feature maps. While the classic network architectures were comprised simply of stacked convolutional layers, modern architectures explore new and innovative ways for constructing convolutional layers in a way which allows for more efficient learning. Almost all of these architectures are based on a repeatable unit which is used throughout the network. These architectures serve as general design guidelines which machine learning practitioners will then adapt to solve various computer vision tasks.

Graphical representation is also very helpful to understand linear algebra. I tried to bind the concepts with plots (and code to produce it). The type of representation I liked most by doing this series is the fact that you can see any matrix as linear transformation of the space. In several chapters we will extend this idea and see how it can be useful to understand eigendecomposition, Singular Value Decomposition (SVD) or the Principal Components Analysis (PCA). In addition, I noticed that creating and reading examples is really helpful to understand the theory. It is why I built Python notebooks.

This post walks through a complete example illustrating an essential data science building block: the underfitting vs. overfitting problem. The author explores the problem through a beginner's implementation of cross-validation. The wide growth of deep learning has complicated things a bit in the hardware department. This post will walk through the different types of computer chips, where they're available, and which ones are the best to boost your performance. One of the most common problems in data science is that of dealing with missing values.

Other hot Machine Learning examples that Jim mentioned were fraud detection, customer service (by understanding customer sentiment and recommending the next best action, e.g. the right person or product), deep insights on asset and supply management, smart cities and drone delivery. And what he really stressed, was that before any model training can take place, data preparation and data organization are critical, pointing to the Hortonworks Data Flow (HDF) and Hortonworks Data Platform (HDP) that together manage the entire data lifecycle, from the edge all the way to the data center, on-prem, in the cloud, or in a hybrid architecture of the two. Only then applying Nvidia's powerful GPUs to train Deep Learning models to drive new insights makes sense.

Global Big Data Conference, the leading vendor agnostic conference for the Big Data (Hadoop, Apache Spark, IoT, Security, NoSQL, Data Science, Machine Learning, Deep Learning, Artificial Intelligence & Predictive Analytics) community, is now announcing its fifth annual event (Aug 28 - Aug 30 2018). The 6th Annual Global Big Data Conference is extended to three days based on feedback from participants. The event will feature many of the Big Data thought leaders from the industry. Annual Global Big Data Conference is an event acclaimed for its highly interactive sessions. Speakers will showcase successful industry vertical use cases, share development and administration tips, and educate organizations about how best to leverage Data (Big Data, Smart Data, Fast Data, Little Data) as a key component in their enterprise data architecture.