The state of the art in video understanding suffers from two problems: (1) The major part of reasoning is performed locally in the video, therefore, it misses important relationships within actions that span several seconds. (2) While there are local methods with fast per-frame processing, the processing of the whole video is not efficient and hampers fast video retrieval or online classification of long-term activities. In this paper, we introduce a network architecture that takes long-term content into account and enables fast per-video processing at the same time. The architecture is based on merging long-term content already in the network rather than in a post-hoc fusion. Together with a sampling strategy, which exploits that neighboring frames are largely redundant, this yields high-quality action classification and video captioning at up to 230 videos per second, where each video can consist of a few hundred frames. The approach achieves competitive performance across all datasets while being 10x to 80x faster than state-of-the-art methods.

Neural Sequence-to-Sequence models have proven to be accurate and robust for many sequence prediction tasks, and have become the standard approach for automatic translation of text. The models work in a five stage blackbox process that involves encoding a source sequence to a vector space and then decoding out to a new target sequence. This process is now standard, but like many deep learning methods remains quite difficult to understand or debug. In this work, we present a visual analysis tool that allows interaction with a trained sequence-to-sequence model through each stage of the translation process. The aim is to identify which patterns have been learned and to detect model errors. We demonstrate the utility of our tool through several real-world large-scale sequence-to-sequence use cases.

These tools don't provide feedback neither about the functional characteristic of a level (e.g. its difficulty and feasibility), nor about the type of player experience a level will enable. Accordingly, designers either perform extensive playtest or build ad-hoc tools to support their content creation activities [1]. In the recent years, artificial intelligence tools have been proposed to assist game designers in the creation of game content like for example the generation of levels that affect players in terms of emotions [2], [3] or that satisfy structural constraints on the position of game elements [4], [5], [6]. In this context, level generation is usually based on design metrics extracted from the level structure (e.g., the position of game elements) or gameplay features (e.g., using data describing players' skill and playing style obtained with sessions of playtesting). In this paper, we present a framework for the design of levels for 2D platformers that extends previous approaches [2], [3] by providing immediate feedback about the functional properties of levels such as (i) the difficulty and probability of success of single jumps (the main mechanic of platformer games), and (ii) a set of statistics to evaluate the difficulty and probability of completion of entire levels. Our framework has been developed as a modular extension of the popular Unity game engine and it is smoothly integrated in its editor.

We study the problem of named entity recognition (NER) from electronic medical records, which is one of the most fundamental and critical problems for medical text mining. Medical records which are written by clinicians from different specialties usually contain quite different terminologies and writing styles. The difference of specialties and the cost of human annotation makes it particularly difficult to train a universal medical NER system. In this paper, we propose a label-aware double transfer learning framework (La-DTL) for cross-specialty NER, so that a medical NER system designed for one specialty could be conveniently applied to another one with minimal annotation efforts. The transferability is guaranteed by 2 components: (i) we propose label-aware MMD for feature representation transfer, and (ii) we perform parameter transfer with a theoretical upper bound which is also label aware. We annotate a new medical NER corpus and conduct extensive experiments on 12 cross-specialty NER tasks. The experimental results demonstrate that La-DTL provides consistent accuracy improvement over strong baselines. Besides, the promising experimental results on non-medical NER scenarios indicate that La-DTL is potential to be seamlessly adapted to a wide range of NER tasks.

We formalize the problem of maximizing the mean-payoff value with high probability while satisfying a parity objective in a Markov decision process (MDP) with unknown probabilistic transition function and unknown reward function. Assuming the support of the unknown transition function and a lower bound on the minimal transition probability are known in advance, we show that in MDPs consisting of a single end component, two combinations of guarantees on the parity and mean-payoff objectives can be achieved depending on how much memory one is willing to use. (i) For all $\epsilon$ and $\gamma$ we can construct an online-learning finite-memory strategy that almost-surely satisfies the parity objective and which achieves an $\epsilon$-optimal mean payoff with probability at least $1 - \gamma$. (ii) Alternatively, for all $\epsilon$ and $\gamma$ there exists an online-learning infinite-memory strategy that satisfies the parity objective surely and which achieves an $\epsilon$-optimal mean payoff with probability at least $1 - \gamma$. We extend the above results to MDPs consisting of more than one end component in a natural way. Finally, we show that the aforementioned guarantees are tight, i.e. there are MDPs for which stronger combinations of the guarantees cannot be ensured.

Quantum walks are the quantum counterpart of classical random walks and, correspondingly, many classical algorithms which are based upon classical random walks naturally lend themselves to be quantized. This quantization, either in the form of coined quantum walks [1-4] or using the so called Szegedy walk operator [5], allows to exploit coherence and entanglement to speedup many computational tasks, with improvements with respect to classical algorithms that are typically quadratic or, in some special circumstances, exponential [6, 7]. Henceforth quantum walks have been employed in a variety of quantum algorithms with applications ranging from statistical physics [8, 9], to combinatorial optimization problems [10], to machine learning [11]. The primary motivation for the quesiton we study in this work is the theory of quantum walks originated from their applicability to quantum machine learning, as demonstrated by the results of [11]. Therein it was shown that it is possible to achieve quadratic speedups in the deliberation time of a learning agent by exploitation of Szegedy quantum walks in conjunction with amplitude amplification. However, together with the certifiable speedups, quantum algorithms usually come together with a plethora of caveats and constraints that can impair their general applicability. In particular, the problem that spurred the present investigation is the following: the algorithm presented in [11] required that a quantum state encoding the stationary distribution is provided at the beginning of the algorithm. While the original paper showed how this assumption can be justified under certain assumptions on the nature of the learning algorithm, the question of other scenarios which provide means to generate such initial states efficiently remained open.

Machine learning is a great way to extract maximum predictive or categorization value from a large volume of structured data. The idea (at least for "supervised learning," by far the most common type in business) is to train a model on a one set of labeled data and then use the resulting models to make predictions or classifications on data where we don't know the outcome. The approach works well in concept, but it can be labor-intensive to develop and deploy the models. One company, however, is rapidly developing a "machine learning machine" that can build and deploy very large numbers of models with relatively little human intervention. You may have heard of dunnhumby, a UK-based analytics company that's owned by the big retailer Tesco.

Deep learning is well known to be very amenable to GPU acceleration. Accelerating "traditional" machine learning methods like logistic regression, linear regression, and support vector machines with GPUs at scale, has, however, been challenging. Today I am very proud to share a major breakthrough that IBM Research has made in this critical area. A team out of our Zurich IBM Research lab beat a previous performance benchmark set for a machine learning workload by Google by 46 times. The research team trained a logistic regression classifier to predict clicks on advertisements using a Terabyte-scale data set that consists of online advertising click-thru data, containing 4.2 billion training examples and 1 million features.

A 14-year-old Everton fan has made history by becoming football's first "remote" match-day mascot - with the aid of a robot. Jack McLinden, who has multiple health conditions, experienced joining his heroes on the pitch before their game against Newcastle United on Monday. Everton captain Phil Jagielka carried the robot, which fed panoramic live images and sound back to Jack's tablet. A microphone enabled him to chat with players in the Goodison Park tunnel. The robot, called AV1, was made by Norwegian firm No Isolation and was designed to help youngsters with long-term illness battle loneliness and attend school from home.

Leading scientists have drawn up plans for a vast multinational European institute devoted to world-class artificial intelligence (AI) research in a desperate bid to nurture and retain top talent in Europe. The new institute would be set up for similar reasons as Cern, the particle physics lab near Geneva, which was created after the second world war to rebuild European physics and reverse the brain drain of the brightest and best scientists to the US. Named the European Lab for Learning and Intelligent Systems, or Ellis, the proposed AI institute would have major centres in a handful of countries, the UK included, with each employing hundreds of computer engineers, mathematicians and other scientists with the express aim of keeping Europe at the forefront of AI research. In an open letter that urges governments to act, the scientists describe how Europe has not kept up with the US and China, where the vast majority of leading AI firms and universities are based. The letter adds that while a few "research hotspots" still exist in Europe, "virtually all of the top people in those places are continuously being pursued for recruitment by US companies."