However, in practice, RNNs can be hard to train and for small to medium-sized training datasets, "good old" methods can often deliver similar or even superior performance at a lower computational cost. Even in the Deep Learning category, RNNs have a strong competitor in Convolutional Neural Nets (a.k.a. ConvNets or CNNs) - just as long as your text can be treated as fixed length sequences, making them a suitable approach to represent and classify tweets, text messages, short user reviews, etc. Still, it's too early to dismiss RNNs and their variants entirely. Where these networks (and particularly their more advanced variant called Long-Short Memory Networks or LSTMs) begin to shine are other NLU tasks that often involve prediction (i.e., generative in nature) rather than "just" classification, a fundamentally discriminative task.

Over the past four years, readers have doubtlessly noticed quantum leaps in the quality of a wide range of everyday technologies. Most obviously, the speech-recognition functions on our smartphones work much better than they used to. When we use a voice command to call our spouses, we reach them now. We aren't connected to Amtrak or an angry ex. In fact, we are increasingly interacting with our computers by just talking to them, whether it's Amazon's Alexa, Apple's Siri, Microsoft's Cortana, or the many voice-responsive features of Google.

Many speeches at AI conferences begin with AlphaGo, the Google-built AI that beat Lee Sedol - one of the highest ranked Go players in the world – to illustrate how far AI has progressed. Most speakers briefly talk about how the computer programme works, and then go on to praise the intelligence of the machine. Oren Etzioni, CEO of the Allen Institute for Artificial Intelligence (AI2), however, gave the credit to the developers and not to AlphaGo. The famous science fiction writer, Arthur C Clarke, once said: "Any sufficiently advanced technology is indistinguishable from magic." "But deep learning is not magic," Etzioni said. "99 per cent of machine learning - or deep learning - is human work. AlphaGo is actually very limited, Etzioni told The Register. "If I asked AlphaGo, can you play poker?

Building Game Bot with OpenAI's Gym (Live Coding starts at 15:00) (youtube.com) I'm really loving what this guy has been doing for making deep learning and machine learning more accessible and fun.

I believe dl4j has been around longer than TF and Torch integration sounds messy since Lua has it's own runtime (there's LuaJ but it looks unmaintained).

Google, Facebook, Amazon, IBM and Microsoft are joining forces to create a new AI partnership dedicated to advancing public understanding of the sector, as well as coming up with standards for future researchers to abide by. Going by the unwieldy name of the Partnership on Artificial Intelligence to Benefit People and Society, the alliance isn't a lobbying organisation (at least, it says it "does not intend" to lobby government bodies). Instead, it says it will "conduct research, recommend best practices, and publish research under an open license in areas such as ethics, fairness and inclusivity; transparency, privacy, and interoperability; collaboration between people and AI systems; and the trustworthiness, reliability and robustness of the technology". There will be equal representation between corporate and non-corporate members on the board of the partnership, and it hopes to invite "academics, non-profits and specialists in policy and ethics" to join. Each of the five founding corporate members has strong AI research teams, some of which have become household names, such as IBM's Watson and Amazon's Alexa.

When it comes to policing artificial intelligence, technology leaders think there is safety in numbers. A new organization called the Partnership on Artificial Intelligence to Benefit People and Society will seek to foster public dialogue and create guidelines for developing AI so that systems do not misbehave. The companies involved include Google and its subsidiary DeepMind, Facebook, Amazon, Microsoft, and IBM. The partnership is founded on eight tenets or principles, including the idea that AI should benefit as many people as possible; that the public should be involved in its development; that research should be conducted in an open way; and that AI systems should be able to explain their reasoning. That such fierce rivals would come together in this way shows how important the companies feel it is to head off public concern and speculation over the potential impacts of AI.

Today a new nonprofit organization called the Partnership on Artificial Intelligence to Benefit People and Society announced its establishment. Participants include Amazon, Google (and Google's DeepMind) Facebook, IBM, and Microsoft. Those companies will put up money and research resources (that could mean talent, open-source code, or data). Company representatives will sit on the organization's board alongside people from academia, the nonprofit world, and policy and ethics experts, according to a statement. Plans call for discussions, studies, reports, best practices, and public outreach in general.

AI is the technology that will have the single biggest impact on cities over the next decade. At Urban.Us, we already meet teams using very large datasets to train algorithms to drive cars, water yards efficiently, guide drones to survey construction sites and route first responders to the people who need them most, and others who use bots to provide legal guidance to people with parking fines. These startups are benefiting from an explosion of data generated by human activities and sensors. Ironically, while the flood of data is difficult for people to understand, it's great for teaching machines. Thanks to cheaper storage and processing to train new algorithms, we've seen a surge in AI deep-learning techniques.

For those curious how NVIDIA's DRIVE PX 2 system is working and how their self-driving car efforts are progressing, they've published a new video today showing their self-driving car that's taught by deep learning with analyzing human driving patterns. NVIDIA's approach to self-driving vehicles is working out well so far and from the deep learning off human driving data is working for driving on unmarked roads, rain conditions, night time driving, around cones, etc. It's also exciting to see how NVIDIA's deep learning efforts have come and that they are Linux powered (Ubuntu still, last I heard from them).