In case you missed it, TensorFlow is now available for Windows, as well as Mac and Linux. This was not always the case. For most of TensorFlow's first year of existence, the only means of Windows support was virtualization, typically through Docker. Even without GPU support, this is great news for me. I teach a graduate course in deep learning and dealing with students who only run Windows was always difficult.

In an age where everything is changing –and changing fast– it's easy to forget how much we've progressed. While we may not have floating cars or robotic teachers, we are on the brink of some very exciting and dramatic developments across all industries. As one of the principal drivers of progression, it's no surprise that learning –and education in general– has been a focus of technological advances. While eLearning is not a new concept, its popularity is increasing, especially as technology becomes more affordable. A big barrier for eLearning is the cost of developing content.

Grasping is a complex process involving knowledge of the object, the surroundings, and of oneself. While humans are able to integrate and process all of the sensory information required for performing this task, equipping machines with this capability is an extremely challenging endeavor. In this paper, we investigate how deep learning techniques can allow us to translate high-level concepts such as motor imagery to the problem of robotic grasp synthesis. We explore a paradigm based on generative models for learning integrated object-action representations, and demonstrate its capacity for capturing and generating multimodal, multi-finger grasp configurations on a simulated grasping dataset.

We solve the compressive sensing problem via convolutional factor analysis, where the convolutional dictionaries are learned in situ from the compressed measurements. An alternating direction method of multipliers (ADMM) paradigm for compressive sensing inversion based on convolutional factor analysis is developed. The proposed algorithm provides reconstructed images as well as features, which can be directly used for recognition (e.g., classification) tasks. When a deep (multilayer) model is constructed, a stochastic unpooling process is employed to build a generative model. During reconstruction and testing, we project the upper layer dictionary to the data level and only a single layer decon-volution is required. We demonstrate that using 30% (relative to pixel numbers) compressed measurements, the proposed model achieves the classification accuracy comparable to the original data on MNIST. We also observe that when the compressed measurements are very limited (e.g., 10%), the upper layer dictionary can provide better reconstruction results than the bottom layer. 1 Introduction The compressive sensing (CS) problem [3,4,11] can be formulated as: min 1 2‖y A x ‖ 2 2 λ‖c ‖?, s.t.x Bc, (1) where A R M N is the sensing matrix and usuallyM N .x is the desired signal, z denotes the coefficients which are sparse ( ‖·‖?

What does artificial intelligence have in common with the price of eggs? Say you're trying to decide between 9 or 10 different varieties of eggs at the store. One catches your eye: "All natural." Well, that's nice, natural is good and they're only 30 cents more -- you buy those. Now, those chickens and the eggs they produce may or may not be more natural than the others -- because there's no official or even generally agreed-upon definition of natural.

How would you feel about getting therapy from a robot? Emotionally intelligent machines may not be as far away as it seems. Over the last few decades, artificial intelligence (AI) have got increasingly good at reading emotional reactions in humans. If AI cannot experience emotions themselves, can they ever truly understand us? And, if not, is there a risk that we ascribe robots properties they don't have? The latest generation of AI's have come about thanks to an increase in data available for computers to learn from, as well as their improved processing power.

The talks at the Deep Learning School on September 24/25, 2016 were amazing. I clipped out individual talks from the full live streams and provided links to each below in case that's useful for people who want to watch specific talks several times (like I do). Please check out the official website (www.bayareadlschool.org)

You may never have heard of diabetic retinopathy, but this nasty condition is the fastest-growing cause of blindness in the world. It poses a risk to the 415 million people with diabetes--nearly 5 percent of the world's population. The condition occurs when chronically high blood sugar damages the tiny vessels that provide blood to the retina. People who suffer from diabetic retinopathy can begin to experience distorted vision and ultimately go blind. And here's the even deeper tragedy: Diabetic retinopathy can be prevented; it just needs to be detected early. With so many people at risk of this condition, the world simply doesn't have enough ophthalmologists available to diagnose them, especially in developing countries.

As we go into 2017 the incipient'technologisation' or'IT-isation', if you'll excuse the terms, of our lives is gathering pace and becoming much plainer to see. AI and deep learning are worth calling out for particular attention. In research consultancy Gartner's'Top 10 Strategic Technology Trends for 2007' survey,[4] Gartner Vice-President and Fellow David Cearley said "over the next 10 years, virtually every app, application and service will incorporate some level of AI. This will form a long-term trend that will continually evolve and expand the application of AI and machine learning for apps and services." Deep learning, a machine learning technique, is emerging as AI's'killer app' enabler.

LinkedIn cofounder Reid Hoffman, Omidyar Network, and John S. and the James L. Knight Foundation have today joined forces to create the Ethics and Governance of Artificial Intelligence Fund. The research fund will focus on AI for the public good and will bring a more diverse range of voices, like faith leaders and policymakers, to AI research and development. Issues the fund may address include ethical design, potential harmful and beneficial impacts of AI, and AI that works in the public interest. "Artificial intelligence and complex algorithms, fueled by big data and deep-learning systems, are quickly changing how we live and work -- from the news stories we see, to the loans for which we qualify, to the jobs we perform," the group said in a statement provided to VentureBeat. "Because of this pervasive but often concealed impact, it is imperative that AI research and development be shaped by a broad range of voices -- not only by engineers and corporations, but also by social scientists, ethicists, philosophers, faith leaders, economists, lawyers and policymakers."