In this paper, we present a simple, yet effective, attention and memory mechanism that is reminiscent of Memory Networks and we demonstrate it in question-answering scenarios. Our mechanism is based on four simple premises: a) memories can be formed from word sequences by using convolutional networks; b) distance measurements can be taken at a neuronal level; c) a recursive softmax function can...

Abstract Representation learning for video is increasingly gaining attention in the field of computer vision. For instance, prediction models enable activity and scene forecasting or vision-based planning control. In this article, we investigate combination differentiable physics spatial transformers a deep action conditional representation network. By our model learns physically interpretable ...

One of the most important fields in robotics is the optimization of controllers. Currently, robots are often treated as a black box in this optimization process, which is the reason why derivative-free optimization methods such as evolutionary algorithms or reinforcement learning are omnipresent. When gradient-based methods are used, models are kept small or rely on finite difference approximat...

Our aim in this paper is twofold. First, to find the necessary and sufficient conditions to be satisfied by a given sequence of real numbers $vn%n50 ` to represent the ‘‘entropic moments’’ * [0,a]@r(x)# dx of an unknown non-negative, decreasing and differentiable ~a.e.! density function r(x) with a finite interval support. These moments are called entropic moments because they are closely conne...

In this paper‎, ‎we introduce and study some new single-valued gap functions for non-differentiable semi-infinite multiobjective optimization problems with locally Lipschitz data‎. ‎Since one of the fundamental properties of gap function for optimization problems is its abilities in characterizing the solutions of the problem in question‎, ‎then the essential properties of the newly introduced ...

This paper presents a complete pipeline for learning continuous motion control policies for a mobile robot when only a non-differentiable physics simulator of robot-terrain interactions is available. The multi-modal state estimation of the robot is also complex and difficult to simulate, so we simultaneously learn a generative model which refines simulator outputs. We propose a coarse-to-fine l...

The problem is from mesoscopic physics: let p : Σ → R3 be an embedded surface in R3, we assume that (1) Σ is orientable, complete, but non-compact; (2) Σ is not totally geodesic; (3) Σ is asymptotically flat in the sense that the second fundamental form goes to zero at infinity. On can build a quantum layer Ω over such a surface Σ as follows: as a differentiable manifold, Ω = Σ × [−a, a] for so...