And other than calculus, linear algebra, optimization, multi variable calculus basic probability and statistics what other advanced math classes should I take?

Google's DeepMind is teaming up with the makers of the StarCraft video game to train its artificial intelligence systems. The AI systems "playing" the game will need to learn strategies similar to those that humans need in the real world, DeepMind said. Its ultimate aim is to develop artificial intelligence that could solve any problem. It has previously taught algorithms to play a range of Atari computer games. StarCraft II, made by developer Blizzard, is a real-time strategy game in which players control one of three warring factions - humans, insects or alien elves.

In a deep learning representation of human disease, lower layers could represent clinical measurements (such as ECG data or protein biomarkers), intermediate layers could represent aberrant pathways (which may simultaneously impact many biomarkers), and top layers could represent disease subclasses (which arise from the variable contributions of 1 aberrant pathways). Ideally, such subclasses would do more than stratify by risk and would actually reflect the dominant disease mechanism(s). This raises a question about the underlying pathophysiologic basis of complex disease in any given individual: is it sparsely encoded in a limited set of aberrant pathways, which could be recovered by an unsupervised learning process (albeit with the right features collected and a large enough sample size), or is it a diffuse, multifactorial process with hundreds of small determinants combining in a highly variable way in different individuals? In the latter case, the concept of precision medicine is unlikely to be of much utility. However, in the former situation, unsupervised and perhaps deep learning might actually realize the elusive goal of reclassifying patients according to more homogenous subgroups, with shared pathophysiology, and the potential of shared response to therapy.

Blizzard made a very curious announcement about Starcraft II at BlizzCon 2016. Instead of a new expansion pack, the game is instead being opened up to Google's Deepmind project; and will teach the AI system how to play an RTS. Deepmind made headlines earlier this year when its AlphaGo AI managed to beat a world class Go player; a feat that was believed to be impossible. The number of possible actions in Go was originally thought to be too great for a computer to calculate within the time constraints of a professional match. Despite this, Deepmind pulled off a 4 – 1 victory over Lee Sedol.

Two crucial questions in discussions about the risks of artificial superintelligence are: 1) How much more capable could an AI become relative to humans, and 2) how easily could superhuman capability be acquired? To answer these questions, I will consider the literature on human expertise and intelligence, discuss its relevance for AI, and consider how an AI could improve on humans in two major aspects of thought and expertise, namely mental simulation and pattern recognition. I find that although there are very real limits to prediction, it seems like an AI could still substantially improve on human intelligence, possibly even mastering domains which are currently too hard for humans. In practice, the limits of prediction do not seem to pose much of a meaningful upper bound on an AI's capabilities, nor do we have any nontrivial lower bounds on how much time it might take to achieve a superhuman level of capability. Takeover scenarios with timescales on the order of mere days or weeks seem to remain within the range of plausibility. As AI systems become more advanced, there is the possibility of them reaching superhuman levels of intelligence, eventually breaking out of human control (Bostrom 2014). The answers to these questions will influence the urgency of dealing with questions of superintelligent AI, as well as the correct means of it. If AI systems can rapidly achieve strong capabilities, becoming powerful enough to take control of the world before any human can react, then that implies a very different approach than one where AI capabilities develop gradually over many decades, never getting substantially past the human level (Sotala & Yampolskiy, 2015). Views on these questions vary. Authors such as Bostrom (2014) and Yudkowsky (2008) argue for the possibility of a fast leap in intelligence, with both offering hypothetical example scenarios where an AI rapidly acquires a dominant position over humanity. On the other hand, Anderson (2010) and Lawrence (2016) appeal to fundamental limits on predictability – and thus intelligence – posed by the complexity of the environment. 'Practitioners who have performed sensitivity analysis on time series prediction will know how quickly uncertainty accumulates as you try to look forward in time. There is normally a time frame ahead of which things become too misty to compute any more. Further computational power doesn't help you in this instance, because uncertainty dominates. Reducing model uncertainty requires exponentially greater computation. We might try to handle this uncertainty by quantifying it, but even this can prove intractable.

Have you ever been in a situation where knowing another language would have come in handy? I remember standing on the platform at Tokyo Station watching my train to Nagano -- the last train of the day -- pulling away without me on it. What ensued was a frustrating hour of gestures, confused smiles, and head-shaking as I wandered the station looking for someone who spoke English (my Japanese is unfortunately nonexistent). It would have been really helpful to have a bilingual pal along with me to translate. Bilingual pals can be hard to find, but Google's new translation software may be an equally useful alternative.

Deep data science is a branch of data science that has little if any overlap with closely related fields such as machine learning, computer science, operations research, mathematics, or statistics. Even classical machine learning and statistical techniques such as clustering, density estimation, or tests of hypotheses, have model-free, data-driven, robust versions designed for automated processing (as in machine-to-machine communications), and thus these techniques also belong to deep data science. Note that unlike deep learning, deep data science is not the intersection of data science and artificial intelligence; however, the analogy between deep data science and deep learning is not completely meaningless, in the sense that both deal with automation. For a robust regression that will work even if all the traditional model assumptions are violated, click here. It is simple (it can be implemented in Excel and it is model-free), efficient and very comparable to the standard regression (when the model assumptions are not violated).

Apache Spark machine learning workloads can run up to 10x faster by moving them to a deep learning paradigm on GPUs, according to Databricks, which today announced that its hosted Spark service on Amazon's new GPU cloud. Databricks, the primary commercial venture behind Apache Spark, today announced that it's now supporting TensorFrames, the new Spark library based on Google's (NASDAQ: GOOG) TensorFlow deep learning framework, on its hosted Spark service, which runs on Amazon Web Services (NASDAQ: AMZM). The deep learning service will be generally available within two weeks, the company says. TensorFrames, which was unveiled this March as a technical preview, lets Spark harness TensorFlow for the purpose of programing deep neural networks, the primary computational method powering so-called "deep learning" algorithms. TensorFrames is also available to on-prem Spark users as a GitHub project, but it's not yet available for download in the Apache Spark project, which limits its usefulness for the time being.

DeepMind, the AI research lab in London that was acquired by Google in 2014 for a reported £400 million, faces one big problem, according to Professor Yann LeCun, who heads up Facebook's AI research group. Notably, LeCun believes that DeepMind, which employs over 250 people and today sits under Alphabet (Google's parent company), is too far away from California. "The challenge I think that DeepMind has is that it's geographically separated from the mothership in California and that makes it very difficult to build technology that can be used in products," LeCun told Business Insider during an interview in London last week. "So it pushes DeepMind to some extent to try to survive on its own." DeepMind declined to comment on this story but it would likely argue that being based in the UK is not a barrier when it comes to working with product and research teams across Google and the rest of the Alphabet group.

In this work, we take a fresh look at some old and new algorithms for off-policy, return-based reinforcement learning. Expressing these in a common form, we derive a novel algorithm, Retrace($\lambda$), with three desired properties: (1) it has low variance; (2) it safely uses samples collected from any behaviour policy, whatever its degree of "off-policyness"; and (3) it is efficient as it makes the best use of samples collected from near on-policy behaviour policies. We analyze the contractive nature of the related operator under both off-policy policy evaluation and control settings and derive online sample-based algorithms. We believe this is the first return-based off-policy control algorithm converging a.s. to $Q^*$ without the GLIE assumption (Greedy in the Limit with Infinite Exploration). As a corollary, we prove the convergence of Watkins' Q($\lambda$), which was an open problem since 1989. We illustrate the benefits of Retrace($\lambda$) on a standard suite of Atari 2600 games.