This paper introduces a cognitive architecture for a humanoid robot to engage in a proactive, mixed-initiative exploration and manipulation of its environment, where the initiative can originate from both the human and the robot. The framework, based on a biologically-grounded theory of the brain and mind, integrates a reactive interaction engine, a number of state-of-the-art perceptual and motor learning algorithms, as well as planning abilities and an autobiographical memory. The architecture as a whole drives the robot behavior to solve the symbol grounding problem, acquire language capabilities, execute goal-oriented behavior, and express a verbal narrative of its own experience in the world. We validate our approach in human-robot interaction experiments with the iCub humanoid robot, showing that the proposed cognitive architecture can be applied in real time within a realistic scenario and that it can be used with naive users.

We consider the problem of non-parametric Conditional Independence testing (CI testing) for continuous random variables. Given i.i.d samples from the joint distribution $f(x,y,z)$ of continuous random vectors $X,Y$ and $Z,$ we determine whether $X \perp Y | Z$. We approach this by converting the conditional independence test into a classification problem. This allows us to harness very powerful classifiers like gradient-boosted trees and deep neural networks. These models can handle complex probability distributions and allow us to perform significantly better compared to the prior state of the art, for high-dimensional CI testing. The main technical challenge in the classification problem is the need for samples from the conditional product distribution $f^{CI}(x,y,z) = f(x|z)f(y|z)f(z)$ -- the joint distribution if and only if $X \perp Y | Z.$ -- when given access only to i.i.d. samples from the true joint distribution $f(x,y,z)$. To tackle this problem we propose a novel nearest neighbor bootstrap procedure and theoretically show that our generated samples are indeed close to $f^{CI}$ in terms of total variational distance. We then develop theoretical results regarding the generalization bounds for classification for our problem, which translate into error bounds for CI testing. We provide a novel analysis of Rademacher type classification bounds in the presence of non-i.i.d near-independent samples. We empirically validate the performance of our algorithm on simulated and real datasets and show performance gains over previous methods.

In this paper, we investigate how to learn to control a group of cooperative agents with limited sensing capabilities such as robot swarms. The agents have only very basic sensor capabilities, yet in a group they can accomplish sophisticated tasks, such as distributed assembly or search and rescue tasks. Learning a policy for a group of agents is difficult due to distributed partial observability of the state. Here, we follow a guided approach where a critic has central access to the global state during learning, which simplifies the policy evaluation problem from a reinforcement learning point of view. For example, we can get the positions of all robots of the swarm using a camera image of a scene. This camera image is only available to the critic and not to the control policies of the robots. We follow an actor-critic approach, where the actors base their decisions only on locally sensed information. In contrast, the critic is learned based on the true global state. Our algorithm uses deep reinforcement learning to approximate both the Q-function and the policy. The performance of the algorithm is evaluated on two tasks with simple simulated 2D agents: 1) finding and maintaining a certain distance to each others and 2) locating a target.

Kernel-based learning algorithms are widely used in machine learning for problems that make use of the similarity between object pairs. Such algorithms first embed all data points into an alternative space, where the inner product between object pairs specifies their distance in the embedding space. Applying kernel methods to partially labeled datasets is a classical challenge in this regard, requiring that the distances between unlabeled pairs must somehow be learnt using the labeled data. In this independent study, I will summarize the work of G. Lanckriet et al.'s work on "Learning the Kernel Matrix with Semidefinite Programming" used in support vector machines (SVM) algorithms for the transduction problem. Throughout the report, I have provide alternative explanations / derivations / analysis related to this work which is designed to ease the understanding of the original article.

We investigate in this paper the architecture of deep convolutional networks. Building on existing state of the art models, we propose a reconfiguration of the model parameters into several parallel branches at the global network level, with each branch being a standalone CNN. We show that this arrangement is an efficient way to significantly reduce the number of parameters without losing performance or to significantly improve the performance with the same level of performance. The use of branches brings an additional form of regularization. In addition to the split into parallel branches, we propose a tighter coupling of these branches by placing the "fuse (averaging) layer" before the Log-Likelihood and SoftMax layers during training. This gives another significant performance improvement, the tighter coupling favouring the learning of better representations, even at the level of the individual branches. We refer to this branched architecture as "coupled ensembles". The approach is very generic and can be applied with almost any DCNN architecture. With coupled ensembles of DenseNet-BC and parameter budget of 25M, we obtain error rates of 2.92%, 15.68% and 1.50% respectively on CIFAR-10, CIFAR-100 and SVHN tasks. For the same budget, DenseNet-BC has error rate of 3.46%, 17.18%, and 1.8% respectively. With ensembles of coupled ensembles, of DenseNet-BC networks, with 50M total parameters, we obtain error rates of 2.72%, 15.13% and 1.42% respectively on these tasks.

Semi-supervised and unsupervised systems provide operators with invaluable support and can tremendously reduce the operators load. In the light of the necessity to process large volumes of video data and provide autonomous decisions, this work proposes new learning algorithms for activity analysis in video. The activities and behaviours are described by a dynamic topic model. Two novel learning algorithms based on the expectation maximisation approach and variational Bayes inference are proposed. Theoretical derivations of the posterior of model parameters are given. The designed learning algorithms are compared with the Gibbs sampling inference scheme introduced earlier in the literature. A detailed comparison of the learning algorithms is presented on real video data. We also propose an anomaly localisation procedure, elegantly embedded in the topic modeling framework. The proposed framework can be applied to a number of areas, including transportation systems, security and surveillance.

In recent years, research in Artificial Intelligence has been primarily dominated by impressive advances in Machine Learning, with a strong emphasis on the so-called Deep Learning framework. It has allowed considerable achievements such as human-level performance in visual classification [1] and description [2], in Atari video games [3] and even in the highly complex game of Go [4]. The Deep Learning approach is characterized by supposing very minimal prior on the task to be solved, compensating this lack of prior knowledge by feeding the learning algorithm with an extremely high amount of training data, while hiding the intermediary representations. However, it is important noting that the most important contributions of Deep Learning for Artificial Intelligence often owe their success in part to their integration with other types of learning algorithms. For example, the AlphaGo program which defeated the world champions in the famously complex game of Go [4], is based on the integration of Deep Reinforcement Learning with a Monte-Carlo tree search algorithm. Without the tree search addition, AlphaGo still outperforms previous machine performances but is unable to beat high-level human players. Another example can be found in the original Deep Q-Learning algorithm (DQN, Mnih et al., 2015), achieving very poor performance in some Atari games where the reward is considerably sparse and delayed (e.g.

Various researchers around the globe are developing ways to detect Alzheimer's as early as possible. After all, early detection gives people the power seek treatment that can slow down the condition's effects, as well as enough time to get their legal and financial affairs in order. Some decided to focus on blood and cerebrospinal fluid tests, while others are developing gadgets that can look for early signs. A team of researchers from the University of Bari in Italy, however, believe the answer lies in artificial intelligence. They developed an algorithm that can spot tiny structural changes in the brain caused by the disease a decade before symptoms even appear.

With this course, you'll have the keys to the entire Python Data Science kingdom!

If you heard Taryn Southern's new single "Break Free" on the radio, you'd probably just keep driving or grocery shopping, or doing whatever you do in places that still have radios playing. The song is a big, moody ballad -- the kind that might play during the climax of a Steven Spielberg movie. "Break Free" wasn't composed by a John Williams copycat, but by artificial intelligence. The song is not a fluke or a novelty for Southern either; she's using artificial intelligence platforms to create an entire album, called I AM AI. It's the first LP to be entirely composed and produced with AI. Southern used an open source AI platform called Amper Music to create the stems of "Break Free."