When we look very closely at images generated by neural networks, we often see a strange checkerboard pattern of artifacts. It’s more obvious in some cases than others, but a large fraction of recent models exhibit this behavior.

Mysteriously, the checkerboard pattern tends to be most prominent in images with strong colors. What’s going on? Do neural networks hate bright colors? The actual cause of these artifacts is actually remarkably simple, as is a method for avoiding them.

Source: Deconvolution and Checkerboard Artifacts — Distill

A popular method for exploring high-dimensional data is something called t-SNE, introduced by van der Maaten and Hinton in 2008. The technique has become widespread in the field of machine learning, since it has an almost magical ability to create compelling two-dimensonal “maps” from data with hundreds or even thousands of dimensions. Although impressive, these images can be tempting to misread. The purpose of this note is to prevent some common misreadings.

Source: How to Use t-SNE Effectively — Distill

Recurrent neural networks are one of the staples of deep learning, allowing neural networks to work with sequences of data like text, audio and video. They can be used to boil a sequence down into a high-level understanding, to annotate sequences, and even to generate new sequences from scratch!

Source: Attention and Augmented Recurrent Neural Networks — Distill

The Deep Learning textbook is a resource intended to help students and practitioners enter the field of machine learning in general and deep learning in particular. The online version of the book is now complete and will remain available online for free.

Source: Deep Learning

Neural Networks and Deep Learning is a free online book. The book will teach you about:

Neural networks and deep learning currently provide the best solutions to many problems in image recognition, speech recognition, and natural language processing. This book will teach you many of the core concepts behind neural networks and deep learning.

Source: Neural networks and deep learning

Several interactive visualizations of a generative model of handwriting. Some are fun, some are serious.

Source: Experiments in Handwriting with a Neural Network — Distill

Embeddings are ubiquitous in machine learning, appearing in recommender systems, NLP, and many other applications. Indeed, in the context of TensorFlow, it’s natural to view tensors (or slices of tensors) as points in space, so almost any TensorFlow system will naturally give rise to various embeddings.

To learn more about embeddings and how to train them, see the Vector Representations of Words tutorial. If you are interested in embeddings of images, check out this article for interesting visualizations of MNIST images. On the other hand, if you are interested in word embeddings, this article gives a good introduction.

TensorBoard has a built-in visualizer, called the Embedding Projector, for interactive visualization and analysis of high-dimensional data like embeddings. It is meant to be useful for developers and researchers alike. It reads from the checkpoint files where you save your tensorflow variables. Although it’s most useful for embeddings, it will load any 2D tensor, potentially including your training weights.

Source: TensorBoard: Embedding Visualization

There’s a projector as well, which you can use seperately from tensorflow here

You can use this to see what your AI is thinking…

DeepMind Lab is a fully 3D game-like platform tailored for agent-based AI research. It is observed from a first-person viewpoint, through the eyes of the simulated agent. Scenes are rendered with rich science fiction-style visuals. The available actions allow agents to look around and move in 3D. The agent’s “body” is a floating orb. It levitates and moves by activating thrusters opposite its desired direction of movement, and it has a camera that moves around the main sphere as a ball-in-socket joint tracking the rotational look actions. Example tasks include collecting fruit, navigating in mazes, traversing dangerous passages while avoiding falling off cliffs, bouncing through space using launch pads to move between platforms, playing laser tag, and quickly learning and remembering random procedurally generated environments.

Source: Open-sourcing DeepMind Lab | DeepMind

github repo here

We’re releasing Universe, a software platform for measuring and training an AI’s general intelligence across the world’s supply of games, websites and other applications.

Universe allows an AI agent to use a computer like a human does: by looking at screen pixels and operating a virtual keyboard and mouse. We must train AI systems on the full range of tasks we expect them to solve, and Universe lets us train a single agent on any task a human can complete with a computer.

In April, we launched Gym, a toolkit for developing and comparing reinforcement learning (RL) algorithms. With Universe, any program can be turned into a Gym environment. Universe works by automatically launching the program behind a VNC remote desktop — it doesn’t need special access to program internals, source code, or bot APIs.

Source: Universe

The homepage
The Git repo

It uses OpenAI Gym for Reinforcement Learning

Reinforcement learning (RL) is the subfield of machine learning concerned with decision making and motor control. It studies how an agent can learn how to achieve goals in a complex, uncertain environment. It’s exciting for two reasons:

RL is very general, encompassing all problems that involve making a sequence of decisions: for example, controlling a robot’s motors so that it’s able to run and jump, making business decisions like pricing and inventory management, or playing video games and board games. RL can even be applied to supervised learning problems with sequential or structured outputs.
RL algorithms have started to achieve good results in many difficult environments. RL has a long history, but until recent advances in deep learning, it required lots of problem-specific engineering. DeepMind’s Atari results, BRETT from Pieter Abbeel’s group, and AlphaGo all used deep RL algorithms which did not make too many assumptions about their environment, and thus can be applied in other settings.

However, RL research is also slowed down by two factors:

The need for better benchmarks. In supervised learning, progress has been driven by large labeled datasets like ImageNet. In RL, the closest equivalent would be a large and diverse collection of environments. However, the existing open-source collections of RL environments don’t have enough variety, and they are often difficult to even set up and use.
Lack of standardization of environments used in publications. Subtle differences in the problem definition, such as the reward function or the set of actions, can drastically alter a task’s difficulty. This issue makes it difficult to reproduce published research and compare results from different papers.

OpenAI Gym is an attempt to fix both problems.

source
The Gym homepage
The Gym github page

“There’s a 60-year-old woman in Tokyo. She was at the University of Tokyo. She had been diagnosed with leukemia six years ago. She was living, but not healthy. So the University of Tokyo ran her genomic sequence through Watson and it was able to ascertain that they were off by one thing. Actually, she had two strains of leukemia. The did treat her and she is healthy.”

He added, “That’s one example. Statistically, we’re seeing that about one third of the time, Watson is proposing an additional diagnosis.”

Source: How IBM Watson saved the life of a woman dying from cancer, exec says – Business Insider

Unfortunately he then goes on to say how great Watson is at pushing ads at you.

AMD is announcing a new series of Radeon-branded products today, targeted at machine intelligence (AI) and deep learning enterprise applications, called Radeon Instinct. As its name suggests, the new Radeon Instinct line of products are comprised of GPU-based solutions for deep learning, inference, and training. The new GPUs are also complemented by a free, open-source library and framework for GPU accelerators, dubbed MIOpen. MIOpen is architected for high-performance machine intelligence applications, and is optimized for the deep learning frameworks in AMD’s ROCm software suite
[…]
The first products in the lineup consist of the Radeon Instinct MI6, the MI8, and the MI25. The 150W Radeon Instinct MI6 accelerator is powered by a Polaris-based GPU, packs 16GB of memory (224GB/s peak bandwidth), and will offer up to 5.7 TFLOPS of peak FP16 performance. Next up in the stack is the Fiji-based Radeon Instinct MI8. Like the Radeon R9 Nano, the Radeon Instinct MI8 features 4GB of High-Bandwidth Memory (HBM), with peak bandwidth of 512GB/s — it’s got a nice small form factor too. The MI8 will offer up to 8.2 TFLOPS of peak FP16 compute performance, with a board power that typical falls below 175W. The Radeon Instinct MI25 accelerator will leverage AMD’s next-generation Vega GPU architecture and has a board power of approximately 300W.

Source: AMD Introduces Radeon Instinct Machine Intelligence And Deep Learning Accelerators

The Chicago Face Database was developed at the University of Chicago by Debbie S. Ma, Joshua Correll, and Bernd Wittenbrink. The CFD is intended for use in scientific research. It provides high-resolution, standardized photographs of male and female faces of varying ethnicity between the ages of 17-65. Extensive norming data are available for each individual model. These data include both physical attributes (e.g., face size) as well as subjective ratings by independent judges (e.g., attractiveness).

Source: Chicago Face Database

Source: China has now eclipsed US in AI / deep-learning research, robotics and VC funding (government-backed $338 billion) · GitHub

TensorFlow™ is an open source software library for numerical computation using data flow graphs. Nodes in the graph represent mathematical operations, while the graph edges represent the multidimensional data arrays (tensors) communicated between them. The flexible architecture allows you to deploy computation to one or more CPUs or GPUs in a desktop, server, or mobile device with a single API. TensorFlow was originally developed by researchers and engineers working on the Google Brain Team within Google’s Machine Intelligence research organization for the purposes of conducting machine learning and deep neural networks research, but the system is general enough to be applicable in a wide variety of other domains as well

Source: TensorFlow — an Open Source Software Library for Machine Intelligence

If you can translate from A to B and from B to C can you translate from A to C without learning the translations directly? Well yes you can. So the translate AI has created its’ own language B (we think) that can function as a midpoint between not only A and C, but also D,E,F, etc.
Would it be what Esperanto wanted to be?

Source: Google’s AI translation tool seems to have invented its own secret internal language | TechCrunch

Using a deep convolutional neural network with residual connections, Miles Deep quickly classifies each second of a pornographic video into 6 categories based on sexual act with 95% accuracy. Then it uses that classification to automatically edit the video. It can remove all the scenes not containing sexual contact, or edit out just a specific act.

Unlike Yahoo’s recently released NSFW model, which uses a similar architecture, Miles Deep can tell the difference between nudity and various explicit sexual acts. As far as I know this is the first and only public pornography classification or editing tool.

Source: GitHub – ryanjay0/miles-deep: Deep Learning Porn Video Classifier/Editor with Caffe

AI Experiments is a showcase for simple experiments that let anyone play with artificial intelligence and machine learning in hands-on ways, through pictures, drawings, language, music, and more.

Source: A.I. Experiments