The NVIDIA Inception Program provides early access to the latest GPU hardware, NVIDIA's deep learning experts and engineering teams, technical training, as well as investment in order help them develop products and services with a first-mover advantage. One of its early collaborations is with NYU, whose researchers are set to work alongside NVIDIA scientists and engineers to develop autonomous driving technology, for which NVIDIA has already created the Drive PX2 chips. The team will seek to grow the the current NVIDIA learning system to encompass all aspects of autonomous driving, eliminating the need for hand-programmed rules and procedures like finding lane markings to avoid the creation of a near infinite number of'if, then, else' statements, which is impractical to code when trying to account for the randomness that occurs on the road.

Remember classic Hollywood films in which someone like Mickey Rooney or Judy Garland gushed, "C'mon, gang, let's get a Long-Short Term Memory (LSTM) Neural Network to put on a show"? Um, well, neither do I; but someday that could be the cliche our children remember. According to CBS News, filmmaker Oscar Sharp and his technologist collaborator Ross Goodwin have built a machine that can write screenplays. They fed the neural network (named Benjamin) hundreds of movie scripts and some prompts and it regurgitated a sparse script, which was turned into the short YouTube video "Sunspring". I'm sure a lot of you assumed that Tinseltown was already using artificial intelligence (AI) to crank out its endless sequels and knuckleheaded knockoffs; but, no, this is a new "thought experiment," as CBS calls it.

I come from the Cambridge machine learning group. More than once I heard people referring to us as "the most Bayesian machine learning group in the world". I mean, we do work with probabilistic models and uncertainty on a daily basis. Maybe that's why it felt so weird playing with those deep learning models (I know, joining the party very late). You see, I spent the last several years working mostly with Gaussian processes, modelling probability distributions over functions. I'm used to uncertainty bounds for decision making, in a similar way many biologists rely on model uncertainty to analyse their data. Working with point estimates alone felt weird to me. I couldn't tell whether the new model I was playing with was making sensible predictions or just guessing at random. I'm certain you've come across this problem yourself, either analysing data or solving some tasks, where you wished you could tell whether your model is certain about its output, asking yourself "maybe I need to use more diverse data? or perhaps change the model?". Most deep learning tools operate in a very different setting to the probabilistic models which possess this invaluable uncertainty information, as one would believe. I recently spent some time trying to understand why these deep learning models work so well – trying to relate them to new research from the last couple of years. I was quite surprised to see how close these were to my beloved Gaussian processes. I was even more surprised to see that we can get uncertainty information from these deep learning models for free – without changing a thing. Update (29/09/2015): I spotted a typo in the calculation of \tau; this has been fixed below.

If TensorFlow is your primary framework, and you are looking for a simple & high-level model definition interface to make your life easier, this tutorial is for you. Keras layers and models are fully compatible with pure-TensorFlow tensors, and as a result, Keras makes a great model definition add-on for TensorFlow, and can even be used alongside other TensorFlow libraries. Note that this tutorial assumes that you have configured Keras to use the TensorFlow backend (instead of Theano). Here are instructions on how to do this. Let's start with a simple example: MNIST digits classification.

Google's DeepMind artificial intelligence (AI) technology has proven to be very smart. DeepMind's AlphaGo system got worldwide attention for beating top-ranked Go player Lee Sedol earlier this year. Previously it has played Breakout and navigated a Doom-like maze. But now the DeepMind software is looking more versatile. Today the Google DeepMind lab unveiled another feat that looks off the wall but is actually evidence of the strength of Google's AI.

Recently I outlined a short series of posts I'll be writing about how I landed a job in autonomous vehicles. My background is that I have a pretty solid foundation in software engineering, including an undergraduate degree in computer science. But most recently my programming has been on the web, not so much in the machine learning and embedded systems areas that dominate vehicle software. Artificial Intelligence for Robotics (Udacity): This is a terrific and super-fun introduction into self-driving cars by Sebastian Thrun. Thrun is both the founder of Udacity and also the founder of Google's self-driving car project and also a former professor at Stanford. Taking the class is like being in the presence of greatness.

Video Friday is your weekly selection of awesome robotics videos, collected by your multilayer Automaton bloggers. We'll also be posting a weekly calendar of upcoming robotics events for the next two months; here's what we have so far (send us your events!): Let us know if you have suggestions for next week, and enjoy today's videos. Also I want that thing that will fire birdies at me. The first robot to autonomously and intentionally break Asimov's first law, which states: A robot may not injure a human being or, through inaction, allow a human being to come to harm.

Machine learning is coming to the data center both to improve internal IT management and embed intelligence into key business processes. You have probably heard of a mystical deep learning, threatening to infuse everything from systems management to self-driving cars. Is this deep learning some really smart artificial intelligence that was just created and about to be unleashed on the world, or simply marketing hype aiming to re-launch complex machine learning algorithms in a better light? It definitely fires the imagination, but it's actually not that complicated. At a technical level, deep learning mostly refers to large compute-intensive neural networks running at scale.

Come on people - let's get our act together on deep learning. I've been studying and writing about DL for close to two years now, and it still amazes the misinformation surrounding this relatively complex learning algorithm. This post is not about how deep learning is or is not overhyped, as that is a well documented debate. This discussion/rant is somewhat off the cuff, but the whole point was to encourage those of us in the machine learning community to think clearly about deep learning. Let's be bold and try to make some claims based on actual science about whether or not this technology will or will not produce artificial intelligence.

Google's week consisted of bringing its TensorFlow machine learning software to iOS, seeing its Tango software added into the first consumer smartphones, and receiving a proposal by its DeepMind division to add a kill switch to artificial intelligence (AI) systems. The firm's AlphaGo AI is also scheduled to face the best Go player the world can offer in a match later this year. Google updated its TensorFlow machine learning software to run on iOS devices. The technology that powers AlphaGo was released to the open source community in November, but has now reached the hands of Apple developers who can build a neural network right into apps. Google's Tango software was revealed to be a key feature of Lenovo's new smartphones.