Attention mechanisms in neural networks are about memory access. That's the first thing to remember about attention: it's something of a misnomer. This YouTube video compares attention mechanism to convolutional networks and recurrent neural networks, explains which problems attention solves better than those other algorithms, and then shows how various attention mechanisms are implemented in Deeplearning4j. "Attention" is defined as the "active direction of the mind to an object", or more literarily as "giving heed".1 The word describes the mind's ability to allocate consideration unevenly across a field of sensation, thought and proprioception, to focus and bring certain inputs to the fore, while ignoring or diminishing the importance of others.

"We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten" -- Bill Gates. When I started working in AI, some 30 years ago, it was virtually unknown -- very few had heard of AI. Nowadays, I rarely meet anyone who has not heard of AI. Hardly a week now passes without major announcements on AI technology – such as that being used in driverless cars, robotics, and other innovate uses. Yet, AI is hardly new. Indeed, the term AI was first coined at the Dartmouth Conference in 1956 – 0ver 60 years ago.

RoboCup Soccer 3D Simulation League (Soccer 3D) is a particularly interesting challenge concerning humanoid robot soccer. It consists of a simulation environment of a soccer match with two teams, each one composed by up to 11 simulated NAO robots [1], the official robot used for RoboCup Standard Platform League since 2008. Soccer 3D is interesting for robotics research since it involves high level multi-agent cooperative decision making while providing a physically realistic environment which requires control and signal processing techniques for robust low level skills. In the current level of evolution of Soccer 3D, motion control is a key factor in team's performance. Indeed, controlling a high degrees of freedom humanoid robot is acknowledged as one of the hardest problems in Robotics. Much effort has been devised to humanoid robot walking, where researchers have been very successful in designing control algorithms which reason about reduced order mathematical models based on the Zero Moment Point (ZMP) concept, such as the linear inverted pendulum model [2]. Nevertheless, these techniques restrict the robot to operate under a small region of its dynamics, where the assumptions of the simplified models are still valid [3, 4]. Therefore, model-based techniques are hard to use for designing highly dynamic movements, such as a long distance kick and a goalkeeper's dive to defend the goal from a fast moving ball.

In this article, we provide and overview of what we consider to be some of the most pressing research questions facing the fields of artificial intelligence (AI) and computational intelligence (CI); with the latter focusing on algorithms that are inspired by various natural phenomena. We demarcate these questions using five unique Rs - namely, (i) rationalizability, (ii) resilience, (iii) reproducibility, (iv) realism, and (v) responsibility. Notably, just as air serves as the basic element of biological life, the term AIR5 - cumulatively referring to the five aforementioned Rs - is introduced herein to mark some of the basic elements of artificial life (supporting the sustained growth of AI and CI). A brief summary of each of the Rs is presented, highlighting their relevance as pillars of future research in this arena.

Deep learning has revolutionized many machine learning tasks in recent years, ranging from image classification and video processing to speech recognition and natural language understanding. The data in these tasks are typically represented in the Euclidean space. However, there is an increasing number of applications where data are generated from non-Euclidean domains and are represented as graphs with complex relationships and interdependency between objects. The complexity of graph data has imposed significant challenges on existing machine learning algorithms. Recently, many studies on extending deep learning approaches for graph data have emerged. In this survey, we provide a comprehensive overview of graph neural networks (GNNs) in data mining and machine learning fields. We propose a new taxonomy to divide the state-of-the-art graph neural networks into different categories. With a focus on graph convolutional networks, we review alternative architectures that have recently been developed; these learning paradigms include graph attention networks, graph autoencoders, graph generative networks, and graph spatial-temporal networks. We further discuss the applications of graph neural networks across various domains and summarize the open source codes and benchmarks of the existing algorithms on different learning tasks. Finally, we propose potential research directions in this fast-growing field.

This study proposes a framework for human-like autonomous car-following planning based on deep reinforcement learning (deep RL). Historical driving data are fed into a simulation environment where an RL agent learns from trial and error interactions based on a reward function that signals how much the agent deviates from the empirical data. Through these interactions, an optimal policy, or car-following model that maps in a human-like way from speed, relative speed between a lead and following vehicle, and inter-vehicle spacing to acceleration of a following vehicle is finally obtained. The model can be continuously updated when more data are fed in. Two thousand car-following periods extracted from the 2015 Shanghai Naturalistic Driving Study were used to train the model and compare its performance with that of traditional and recent data-driven car-following models. As shown by this study results, a deep deterministic policy gradient car-following model that uses disparity between simulated and observed speed as the reward function and considers a reaction delay of 1s, denoted as DDPGvRT, can reproduce human-like car-following behavior with higher accuracy than traditional and recent data-driven car-following models. Specifically, the DDPGvRT model has a spacing validation error of 18% and speed validation error of 5%, which are less than those of other models, including the intelligent driver model, models based on locally weighted regression, and conventional neural network-based models. Moreover, the DDPGvRT demonstrates good capability of generalization to various driving situations and can adapt to different drivers by continuously learning. This study demonstrates that reinforcement learning methodology can offer insight into driver behavior and can contribute to the development of human-like autonomous driving algorithms and traffic-flow models.

This work presents a new strategy for multi-class classification that requires no class-specific labels, but instead leverages pairwise similarity between examples, which is a weaker form of annotation. The proposed method, meta classification learning, optimizes a binary classifier for pairwise similarity prediction and through this process learns a multi-class classifier as a submodule. We formulate this approach, present a probabilistic graphical model for it, and derive a surprisingly simple loss function that can be used to learn neural network-based models. We then demonstrate that this same framework generalizes to the supervised, unsupervised cross-task, and semi-supervised settings. Our method is evaluated against state of the art in all three learning paradigms and shows a superior or comparable accuracy, providing evidence that learning multi-class classification without multi-class labels is a viable learning option.

Cosmologists are facing the problem of the analysis of a huge quantity of data when observing the sky. The methods used in cosmology are, for the most of them, relying on astrophysical models, and thus, for the classification, they usually use a machine learning approach in two-steps, which consists in, first, extracting features, and second, using a classifier. In this paper, we are specifically studying the supernovae phenomenon and especially the binary classification "I.a supernovae versus not-I.a supernovae". We present two Convolutional Neural Networks (CNNs) defeating the current state-of-the-art. The first one is adapted to time series and thus to the treatment of supernovae light-curves. The second one is based on a Siamese CNN and is suited to the nature of data, i.e. their sparsity and their weak quantity (small learning database).

Stochastic gradient descent (SGD) has been found to be surprisingly effective in training a variety of deep neural networks. However, there is still a lack of understanding on how and why SGD can train these complex networks towards a global minimum. In this study, we establish the convergence of SGD to a global minimum for nonconvex optimization problems that are commonly encountered in neural network training. Our argument exploits the following two important properties: 1) the training loss can achieve zero value (approximately), which has been widely observed in deep learning; 2) SGD follows a star-convex path, which is verified by various experiments in this paper. In such a context, our analysis shows that SGD, although has long been considered as a randomized algorithm, converges in an intrinsically deterministic manner to a global minimum.

A long standing open problem in the theory of neural networks is the development of quantitative methods to estimate and compare the capabilities of different architectures. Here we define the capacity of an architecture by the binary logarithm of the number of functions it can compute, as the synaptic weights are varied. The capacity is an upper bound on the number of bits that can be "communicated" from the training data to the architecture over the learning channel. We study the capacity of layered, fully-connected, architectures of linear threshold neurons with $L$ layers of size $n_1,n_2, \ldots, n_L$ and show that in essence the capacity is given by a cubic polynomial in the layer sizes: $C(n_1,\ldots, n_L)=\sum_{k=1}^{L-1} \min(n_1,\ldots,n_k)n_kn_{k+1}$. In proving the main result, we also develop new techniques (multiplexing, enrichment, and stacking) as well as new bounds on the capacity of finite sets. We use the main result to identify architectures with maximal or minimal capacity under a number of natural constraints. This leads to the notion of structural regularization for deep architectures. While in general, everything else being equal, shallow networks compute more functions than deep networks, the functions computed by deep networks are more regular and "interesting".