Demis Hassabis knows a thing or two about artificial intelligence: he founded the London-based AI startup DeepMind, which was purchased by Google for $650 million back in 2014. Since then, his company has wiped the floor with humans at the complex game of Go and begun making steps towards crafting more general AIs. But now he's come out and said that be believes the only way for artificial intelligence to realize its true potential is with a dose of inspiration from human intellect. Currently, most AI systems are based on layers of mathematics that are only loosely inspired by the way the human brain works. But different types of machine learning, such as speech recognition or identifying objects in an image, require different mathematical structures, and the resulting algorithms are only able to perform very specific tasks.

Summary Full-batch GD SGD Momentum SGD NAG AdaGrad RMSProp Adam Speed up by momentum Adaptive learning rate 30. More deep learning coming up! • Optimization in Deep learning (today's session) • Behind AlphaGo • Mastering the game of Go with deep neural networks and tree search • Attention network • Application of Deep Learning and showcase • Any proposal?

Global Deep Learning Market size is expected to reach USD 10.2 billion by 2025, according to a new report by Grand View Research, Inc. Considerable improvements in machine learning algorithms and advancements in deep learning chipsets are driving the industry growth. Rapid improvements in fast information storage capacity, high computing power, and parallelization have contributed to the swift uptake of the deep learning technology in end-use industries such as automotive and healthcare. Further, the need for understanding and analyzing visual contents among enterprises in order to gain meaningful insights, is expected to provide traction to the industry over the forecast period. The increasing prominence of Graphics Processing Unit (GPU)-accelerated applications is leading to increased adoption of the technology in scientific disciplines such as deep learning and data science. Organizations are utilizing deep learning neural networks to extract valuable insights from enormous amounts of data for providing innovative products and improving customer experience; thereby, increasing revenue opportunities.

Resolving abstract anaphora is an important, but difficult task for text understanding. Yet, with recent advances in representation learning this task becomes a more tangible aim. A central property of abstract anaphora is that it establishes a relation between the anaphor embedded in the anaphoric sentence and its (typically non-nominal) antecedent. We propose a mention-ranking model that learns how abstract anaphors relate to their antecedents with an LSTM-Siamese Net. We overcome the lack of training data by generating artificial anaphoric sentence--antecedent pairs. Our model outperforms state-of-the-art results on shell noun resolution. We also report first benchmark results on an abstract anaphora subset of the ARRAU corpus. This corpus presents a greater challenge due to a mixture of nominal and pronominal anaphors and a greater range of confounders. We found model variants that outperform the baselines for nominal anaphors, without training on individual anaphor data, but still lag behind for pronominal anaphors. Our model selects syntactically plausible candidates and -- if disregarding syntax -- discriminates candidates using deeper features.

We present a learning-based mapless motion planner by taking the sparse 10-dimensional range findings and the target position with respect to the mobile robot coordinate frame as input and the continuous steering commands as output. Traditional motion planners for mobile ground robots with a laser range sensor mostly depend on the obstacle map of the navigation environment where both the highly precise laser sensor and the obstacle map building work of the environment are indispensable. We show that, through an asynchronous deep reinforcement learning method, a mapless motion planner can be trained end-to-end without any manually designed features and prior demonstrations. The trained planner can be directly applied in unseen virtual and real environments. The experiments show that the proposed mapless motion planner can navigate the nonholonomic mobile robot to the desired targets without colliding with any obstacles.

We propose a method for learning expressive energy-based policies for continuous states and actions, which has been feasible only in tabular domains before. We apply our method to learning maximum entropy policies, resulting into a new algorithm, called soft Q-learning, that expresses the optimal policy via a Boltzmann distribution. We use the recently proposed amortized Stein variational gradient descent to learn a stochastic sampling network that approximates samples from this distribution. The benefits of the proposed algorithm include improved exploration and compositionality that allows transferring skills between tasks, which we confirm in simulated experiments with swimming and walking robots. We also draw a connection to actor-critic methods, which can be viewed performing approximate inference on the corresponding energy-based model.

Over the past few months, I have been collecting AI cheat sheets. From time to time I share them with friends and colleagues and recently I have been getting asked a lot, so I decided to organize and share the entire collection. To make things more interesting and give context, I added descriptions and/or excerpts for each major topic.

Motivation: Text mining has become an important tool for biomedical research. The most fundamental text-mining task is the recognition of biomedical named entities (NER), such as genes, chemicals and diseases. Current NER methods rely on pre-defined features which try to capture the specific surface properties of entity types, properties of the typical local context, background knowledge, and linguistic information. State-of-the-art tools are entity-specific, as dictionaries and empirically optimal feature sets differ between entity types, which makes their development costly. Furthermore, features are often optimized for a specific gold standard corpus, which makes extrapolation of quality measures difficult. Results: We show that a completely generic method based on deep learning and statistical word embeddings [called long short-term memory network-conditional random field (LSTM-CRF)] outperforms state-of-the-art entity-specific NER tools, and often by a large margin. To this end, we compared the performance of LSTM-CRF on 33 data sets covering five different entity classes with that of best-of-class NER tools and an entity-agnostic CRF implementation. On average, F1-score of LSTM-CRF is 5% above that of the baselines, mostly due to a sharp increase in recall. Availability and implementation: The source code for LSTM-CRF is available at https://github.com/glample/tagger Text mining is an important tool for many types of large-scale biomedical data analysis, such as network biology (Zhou et al., 2014), gene prioritization (Aerts et al., 2006), drug repositioning (Wang and Zhang, 2013) or creation of curated databases (Li et al., 2015).

Natural Language Processing (NLP) of texts has been applied with different degrees of success. For example, automatic translation has attracted a lot of attention in the early stages of NLP. Nowadays, with the advent of social networks, users generate a big volume of interesting information for companies which are either in the search of user feedback for they products or in the search of personalised information to sell new ones. Thus, new NLP interesting applications appear such as sentiment analysis (extracting opinions in a user opinion about a product), user wants and needs detection or user profiling. Humans cannot process this information timely without great effort and money expenditures and computers stand up as the only alternative as they are much faster than humans.

In the latest sign of artificial intelligence (AI)'s eventual dominance of the workplace, a Canadian deep learning startup-turned-division of Microsoft Corp. has successfully created an AI-based system that achieved the maximum possible score on Ms. Pac-Man. That might not sound like the most complicated task in the world – especially since the edition in question was the Atari 2600 version and not the arcade original – but as Microsoft senior writer Allison Linn explains in a recent blog post, the challenge facing researchers at Montreal-based Maluuba was more daunting than you might think. "A lot of companies working on AI use games to build intelligent algorithms because there's a lot of human-like intelligence capabilities that you need to beat the games," Maluuba program manager Rahul Mehrotra explains in the story, noting that the variety of situations you can encounter while playing the games makes them a good testing ground. In other words, the techniques used to develop the AI-driven Ms. Pac-Man master (or is that mistress?) Like many of its ilk, Ms. Pac-Man was intentionally designed to be easy to learn yet nearly impossible to master so that players would keep dropping in quarters, with co-creator Steve Golson noting that Ms. Pac-Man in particular was programmed to be more random than the original Pac-Man, so it would be harder for players to finish.