An assumption-free automatic check of medical images for potentially overseen anomalies would be a valuable assistance for a radiologist. Deep learning and especially Variational Auto-Encoders (VAEs) have shown great potential in the unsupervised learning of data distributions. In principle, this allows for such a check and even the localization of parts in the image that are most suspicious. Currently, however, the reconstruction-based localization by design requires adjusting the model architecture to the specific problem looked at during evaluation. This contradicts the principle of building assumption-free models. We propose complementing the localization part with a term derived from the Kullback-Leibler (KL)-divergence. For validation, we perform a series of experiments on FashionMNIST as well as on a medical task including >1000 healthy and >250 brain tumor patients. Results show that the proposed formalism outperforms the state of the art VAE-based localization of anomalies across many hyperparameter settings and also shows a competitive max performance.

The advance of node pooling operations in Graph Neural Networks (GNNs) has lagged behind the feverish design of new message-passing techniques, and pooling remains an important and challenging endeavor for the design of deep architectures. In this paper, we propose a pooling operation for GNNs that leverages a differentiable unsupervised loss based on the mincut optimization objective. For each node, our method learns a soft cluster assignment vector that depends on the node features, the target inference task (e.g., graph classification), and, thanks to the mincut objective, also on the graph connectivity. Graph pooling is obtained by applying the matrix of assignment vectors to the adjacency matrix and the node features. We validate the effectiveness of the proposed pooling method on a variety of supervised and unsupervised tasks.

Plotting a learner's average performance against the number of training samples results in a learning curve. Studying such curves on one or more data sets is a way to get to a better understanding of the generalization properties of this learner. The behavior of learning curves is, however, not very well understood and can display (for most researchers) quite unexpected behavior. Our work introduces the formal notion of \emph{risk monotonicity}, which asks the risk to not deteriorate with increasing training set sizes in expectation over the training samples. We then present the surprising result that various standard learners, specifically those that minimize the empirical risk, can act \emph{non}monotonically irrespective of the training sample size. We provide a theoretical underpinning for specific instantiations from classification, regression, and density estimation. Altogether, the proposed monotonicity notion opens up a whole new direction of research.

The complexity, heterogeneity and scale of electrical networks have grown far beyond the limits of exclusively human-based management at the Smart Grid (SG). Likewise, researchers cogitate the use of artificial intelligence and heuristics techniques to create cognitive and autonomic management tools that aim better assist and enhance SG management processes like in the grid reconfiguration. The development of self-healing management approaches towards a cognitive and autonomic distribution power network reconfiguration is a scenario in which the scalability and on-the-fly computation are issues. This paper proposes the use of Case-Based Reasoning (CBR) coupled with the HATSGA algorithm for the fast reconfiguration of large distribution power networks. The suitability and the scalability of the CBR-based reconfiguration strategy using HATSGA algorithm are evaluated. The evaluation indicates that the adopted HATSGA algorithm computes new reconfiguration topologies with a feasible computational time for large networks. The CBR strategy looks for managerial acceptable reconfiguration solutions at the CBR database and, as such, contributes to reduce the required number of reconfiguration computation using HATSGA. This suggests CBR can be applied with a fast reconfiguration algorithm resulting in more efficient, dynamic and cognitive grid recovery strategy.

Despite their accuracy, neural network-based classifiers are still prone to manipulation through adversarial perturbations. Those perturbations are designed to be misclassified by the neural network, while being perceptually identical to some valid input. The vast majority of attack methods rely on white-box conditions, where the attacker has full knowledge of the attacked network's parameters. This allows the attacker to calculate the network's loss gradient with respect to some valid input and use this gradient in order to create an adversarial example. The task of blocking white-box attacks has proven difficult to solve. While a large number of defense methods have been suggested, they have had limited success. In this work we examine this difficulty and try to understand it. We systematically explore the abilities and limitations of defensive distillation, one of the most promising defense mechanisms against adversarial perturbations suggested so far in order to understand the defense challenge. We show that contrary to commonly held belief, the ability to bypass defensive distillation is not dependent on an attack's level of sophistication. In fact, simple approaches, such as the Targeted Gradient Sign Method, are capable of effectively bypassing defensive distillation. We prove that defensive distillation is highly effective against non-targeted attacks but is unsuitable for targeted attacks. This discovery leads us to realize that targeted attacks leverage the same input gradient that allows a network to be trained. This implies that blocking them will require losing the network's ability to learn, presenting an impossible tradeoff to the research community.

Many real-world human behaviors can be characterized as a sequential decision making processes, such as urban travelers choices of transport modes and routes (Wu et al. 2017). Differing from choices controlled by machines, which in general follows perfect rationality to adopt the policy with the highest reward, studies have revealed that human agents make sub-optimal decisions under bounded rationality (Tao, Rohde, and Corcoran 2014). Such behaviors can be modeled using maximum causal entropy (MCE) principle (Ziebart 2010). In this paper, we define and investigate a general reward trans-formation problem (namely, reward advancement): Recovering the range of additional reward functions that transform the agent's policy from original policy to a predefined target policy under MCE principle. We show that given an MDP and a target policy, there are infinite many additional reward functions that can achieve the desired policy transformation. Moreover, we propose an algorithm to further extract the additional rewards with minimum "cost" to implement the policy transformation.

This paper introduces a new perspective of intelligent robots and systems control. The presented and proposed cognitive model: Memory, Learning and Recognition (MLR), is an effort to bridge the gap between Robotics, AI, Cognitive Science, and Neuroscience. The currently existing gap prevents us from integrating the current advancement and achievements of these four research fields which are actively trying to define intelligence in either application-based way or in generic way. This cognitive model defines intelligence more specifically, parametrically and detailed. The proposed MLR model helps us create a general control model for robots and systems independent of their application domains and platforms since it is mainly based on the dataset provided for robots and systems controls. This paper is mainly proposing and introducing this concept and trying to prove this concept in a small scale, firstly through experimentation. The proposed concept is also applicable to other different platforms in real-time as well as in simulation.

The Viral Marketing is a relatively new form of marketing that exploits social networks to promote a brand, a product, etc. The idea behind it is to find a set of influencers on the network that can trigger a large cascade of propagation and adoptions. In this paper, we will introduce an evidential opinion-based influence maximization model for viral marketing. Besides, our approach tackles three opinions based scenarios for viral marketing in the real world. The first scenario concerns influencers who have a positive opinion about the product. The second scenario deals with influencers who have a positive opinion about the product and produce effects on users who also have a positive opinion. The third scenario involves influence users who have a positive opinion about the product and produce effects on the negative opinion of other users concerning the product in question. Next, we proposed six influence measures, two for each scenario. We also use an influence maximization model that the set of detected influencers for each scenario. Finally, we show the performance of the proposed model with each influence measure through some experiments conducted on a generated dataset and a real world dataset collected from Twitter.

Cooperative game is a critical research area in multi-agent reinforcement learning (MARL). Global reward game is a subclass of cooperative games, where all agents aim to maximize cumulative global rewards. Credit assignment is an important problem studied in the global reward game. Most works stand by the view of non-cooperative-game theoretical framework with the shared reward approach, i.e., each agent is assigned a shared global reward directly. This, however, may give each agent an inaccurate feedback on his contribution to the group. In this paper, we introduce a cooperative-game theoretical framework and extend it to the finite-horizon case. We show that our proposed framework is a superset of the global reward game. Based on this framework, we propose an algorithm called Shapley Q-value policy gradient (SQPG) to learn a local reward approach that can distribute the cumulative global reward fairly, reflecting each agent's own contribution in contrast to the shared reward approach. We evaluate our method on the Cooperative Navigation, Prey-and-Predator and Traffic Junction, compared with MADDPG, COMA, Independent actor-critic and Independent DDPG. In the experiments, our algorithm shows better convergence than the baselines.

Multi-Agent Motion Planning (MAMP) is the task of finding conflict-free kinodynamically feasible plans for agents from start to goal states. While MAMP is of significant practical importance, existing solvers are either incomplete, inefficient or rely on simplifying assumptions. For example, Multi-Agent Path Finding (MAPF) solvers conventionally assume discrete timesteps and rectilinear movement of agents between neighboring vertices of a graph. In this paper, we develop MAMP solvers that obviate these simplifying assumptions and yet generalize the core ideas of state-of-the-art MAPF solvers. Specifically, since different motions may take arbitrarily different durations, MAMP solvers need to efficiently reason with continuous time and arbitrary wait durations. To do so, we adapt (Enhanced) Conflict-Based Search to continuous time and develop a novel bounded-suboptimal extension of Safe Interval Path Planning, called Soft Conflict Interval Path Planning. On the theoretical side, we justify the completeness, optimality and bounded-suboptimality of our MAMP solvers. On the experimental side, we show that our MAMP solvers scale well with increasing suboptimality bounds.