查看更多>>摘要：This paper reviews exploration techniques in deep reinforcement learning. Exploration techniques are of primary importance when solving sparse reward problems. In sparse reward problems, the reward is rare, which means that the agent will not find the reward often by acting randomly. In such a scenario, it is challenging for reinforcement learning to learn rewards and actions association. Thus more sophisticated exploration methods need to be devised. This review provides a comprehensive overview of existing exploration approaches, which are categorised based on the key contributions as: reward novel states, reward diverse behaviours, goal-based methods, probabilistic methods, imitation-based methods, safe exploration and random-based methods. Then, unsolved challenges are discussed to provide valuable future research directions. Finally, the approaches of different categories are compared in terms of complexity, computational effort and overall performance.
查看更多>>摘要：Observability is a fundamental structural property of any dynamic system and describes the possibility of reconstructing the state that characterizes the system from fusing the observations of its inputs and outputs. Despite the effort made to study this property and to introduce analytical criteria capable of verifying whether or not a dynamic system satisfies this property, there is no general analytical criterion to obtain the observability of the state when the dynamics are also driven by unknown inputs. Here, we introduce the general analytical solution of this fundamental open problem, often called the unknown input observability problem. We provide the systematic procedure, based on automatic calculation (differentiation and determination of the matrix rank), which allows us to check the observability of the state even in the presence of unknown inputs. One of the fundamental ingredients to obtain this solution is the characterization of the group of invariance of observability. We have very recently introduced this group, together with a new set of tensor fields with respect to this group of transformations (Martinelli, 2020). The analytical solution of the unknown input observability problem is expressed in terms of these tensor fields. In Martinelli (2020) we provided the solution by restricting our investigation to systems that satisfy a special assumption that is called canonicity with respect to the unknown inputs. Here, after an exhaustive characterization of the concept of canonicity, we also account for the case when this assumption is not satisfied and we provide the general solution. This solution is also provided in the form of a new algorithm. In addition, even in the canonic case dealt with in Martinelli (2020), here we provide a new fundamental result that regards the convergence properties of the solution. Finally, as a consequence of the results obtained here, we also provide the condition to reconstruct the unknown inputs, and, when this condition is not met, what can be reconstructed on the unknown inputs. We illustrate the implementation of the new algorithm by studying the observability properties of a nonlinear system in the framework of visual-inertial sensor fusion, whose dynamics are driven by two unknown inputs and one known input. In particular, for this system, we follow step by step the algorithm introduced by this paper, which solves the unknown input observability problem in the most general case.
查看更多>>摘要：Event classification is inherently sequential and multimodal. Therefore, deep neural models need to dynamically focus on the most relevant time window and/or modality of a video. In this study, we propose the Multimodal Attentive Fusion Network (MAFnet), an architecture that can dynamically fuse visual and audio information for event recognition. Inspired by prior studies in neuroscience, we couple both modalities at different levels of visual and audio paths. Furthermore, the network dynamically highlights a modality at a given time window relevant to classify events. Experimental results in AVE (Audio-Visual Event), UCF51, and Kinetics-Sounds datasets show that the approach can effectively improve the accuracy in audio-visual event classification. Code is available at: https://github.com/numediart/MAFnet
查看更多>>摘要：Guided image denoising recovers clean target images by fusing guidance images and noisy target images. Several deep neural networks have been designed for this task, but they are black-box methods lacking interpretability. To overcome the issue, this paper builds a more interpretable network. To start with, an observation model is proposed to account for modality gap between target and guidance images. Then, this paper formulates a deep prior regularized optimization problem, and solves it by alternating direction method of multipliers (ADMM) algorithm. The update rules are generalized to design the network architecture. Extensive experiments conducted on FAIP and RNS datasets manifest that the novel network outperforms several state-of-the-art and benchmark methods regarding both evaluation metrics and visual inspection.
NSTL
Elsevier