The following interview is one of many included in the report. As part of our ongoing series of interviews surveying the frontiers of machine intelligence, I recently interviewed Brendan Frey. Frey is a co-founder of Deep Genomics, a professor at the University of Toronto and a co-founder of its Machine Learning Group, a senior fellow of the Neural Computation program at the Canadian Institute for Advanced Research, and a fellow of the Royal Society of Canada. His work focuses on using machine learning to understand the genome and to realize new possibilities in genomic medicine. Brendan Frey: I completed my Ph.D. with Geoff Hinton in 1997.

Last December, Tesla CEO Elon Musk teamed up with Y Combinator president Sam Altman and former Google Brain Team scientist Ilya Sutskever to launch OpenAI, a 1 billion non-profit organization dedicated to furthering our understanding of artificial intelligence with a promise to share its research openly with the world. Today, it's taken its first step in that direction by launching a free toolkit for developers to build and train their own AI bots with games and algorithmic challenges. Some of the biggest names in tech are coming to TNW Conference in Amsterdam this May. The OpenAI Gym, currently in beta, includes environments to simulate situations for your AI to learn from, as well as a site to compare and reproduce results. The tools are designed for use with Reinforcement Learning (RL), one of the technologies used to develop Google's AlphaGo AI that defeated Go world champion Lee Se-Dol recently. RL works on the principle that a bot will receive a reward every time it completes an action successfully – similar to how you might train a dog.

If you're looking to add artificial intelligence to a hardware project for doing things like sensing objects in an enviroment or understand voice commands, Movidius' new Fathom USB stick might be just the thing. The company is known for its Myriad 2 deep learning chip that allows DJI drones to avoid obstacles. The Fathom is essentially a portable version of the chip that can be plugged into the USB 3.0 port of Linux-based devices to run fully-trained neural networks while consuming very little power. Some of the biggest names in tech are coming to TNW Conference in Amsterdam this May. It's compatible with Caffe and TensorFlow frameworks and is capable of 150 gigaFLOPS (150 billion floating-operations per second). That means developers can use it to do things like enable robots to understand natural human speech and recognize faces, and teach drones to navigate indoors and outdoors by themselves – all without the need to connect to the cloud.

OpenAI, a 1 billion ( 687 million) artificial intelligence company backed by Elon Musk, has built a "gym" where developers can train their AI systems to get smarter. Using OpenAI's open source toolkit, available for download now, developers can access "environments" where they can test their AI bots. The OpenAI Gym, currently in beta, provides a number of environments, including more than 50 Atari games, such as "Space Invaders," "Pong," "Asteroids" and "Pac-Man". Developers can also test their AIs on board games like Go, which was recently mastered by an agent built by London startup Google DeepMind. "Over time, we plan to greatly expand this collection of environments," wrote OpenAI's Greg Brockman and John Schulman in a blog post.

Irish chip maker Movidius has created the world's first deep learning USB stick that can add artificial intelligence (AI) to future products from self-driving cars to robots, and drones that will learn to think for themselves. Entitled the Fathom Neural Compute Stick, the device will sell for less than 100 and will allow powerful neural networks to be moved out of the cloud and deployed on new products like robots and drones. It is the latest breakthrough for the Dublin company, which has been winning major multi-million dollar deals with Google and drone maker DJI. 'With Fathom, every robot, big and small, can now have state-of-the-art vision capabilities' – DR YANN LECUN, NEW YORK UNIVERSITY "Any organisation can now add deep learning or machine intelligence to devices using the USB stick and create products that will be accessible to broader markets," Movidius co-founder David Moloney told Siliconrepublic.com. "We've already seen how the auto industry has been outflanked by Tesla ...

Elon Musk's OpenAI has created a'gym' to let developers train their AI systems on Atari games. The open source code, which is still in development, includes'environments' to create situations in which AI can learn. The environments include playing classic board games, controlling a robot in simulation and playing 59 Atari games like Asteroids, Air Raid, Pac-Man, Space Invaders and Pitfall. The hope is that the tasks will give OpenAI and others a way to rank and improve various AI approaches, and unveil new ways to teach machines to learn. OpenAI will also feature a leaderboard of the most successful systems.

Tech billionaire Elon Musk's artificial intelligence company has opened up a virtual'gym' to enable developers to train their AIs using vintage video games like Pac-Man. OpenAI launched the open source code workout spot in order to offer'environments' in which tech boffins can test their AIs. The environments on offer include a range of 59 classic Atari games such as Pong and Asteroids. The code-based gym also includes the strategy board game Go. Google's DeepMind AI software recently beat the reigning human Go champion in a series of matches.

Tesla Motors cofounder and SpaceX CEO Elon Musk recently collaborated with Y Combinator's Sam Altman and former Googler Ilya Sutskever to launch a non-profit platform for artificial intelligence research called OpenAI. Artificial intelligence (AI) is no longer just a part of fictional films today. As a matter of fact, AI has become ubiquitous in almost all fields of sciences, offering its remarkable benefits that can revolutionize the world. Despite its protracted history, artificial Intelligence is a field that is still constantly and actively evolving. It is defined as the science and engineering of making intelligent machines through incorporating clever computer programs.

Human activity recognition (HAR) in ubiquitous computing is beginning to adopt deep learning to substitute for well-established analysis techniques that rely on hand-crafted feature extraction and classification techniques. From these isolated applications of custom deep architectures it is, however, difficult to gain an overview of their suitability for problems ranging from the recognition of manipulative gestures to the segmentation and identification of physical activities like running or ascending stairs. In this paper we rigorously explore deep, convolutional, and recurrent approaches across three representative datasets that contain movement data captured with wearable sensors. We describe how to train recurrent approaches in this setting, introduce a novel regularisation approach, and illustrate how they outperform the state-of-the-art on a large benchmark dataset. Across thousands of recognition experiments with randomly sampled model configurations we investigate the suitability of each model for different tasks in HAR, explore the impact of hyperparameters using the fANOVA framework, and provide guidelines for the practitioner who wants to apply deep learning in their problem setting.

We introduce a simple recurrent variational auto-encoder architecture that significantly improves image modeling. The system represents the state-of-the-art in latent variable models for both the ImageNet and Omniglot datasets. We show that it naturally separates global conceptual information from lower level details, thus addressing one of the fundamentally desired properties of unsupervised learning. Furthermore, the possibility of restricting ourselves to storing only global information about an image allows us to achieve high quality 'conceptual compression'.