A5030762555- Reinforcement Learning in Robotics
- Algebraic structures and combinatorial models
- Efficiency Analysis Using DEA
- Adaptive Dynamic Programming Control
- Advanced Algebra and Geometry
- Multi-Criteria Decision Making
- Environmental Impact and Sustainability
- Domain Adaptation and Few-Shot Learning
- Evolutionary Algorithms and Applications
- Nonlinear Waves and Solitons
- Elevator Systems and Control
- Energy, Environment, Economic Growth
- Optimization and Mathematical Programming
- Economic Growth and Productivity
- Neural Networks and Reservoir Computing
- Advanced Combinatorial Mathematics
- Artificial Intelligence in Healthcare
- Commutative Algebra and Its Applications
- Advanced Bandit Algorithms Research
- Fuel Cells and Related Materials
- Autophagy in Disease and Therapy
- Advanced Statistical Methods and Models
- HIV Research and Treatment
- Imbalanced Data Classification Techniques
- Robot Manipulation and Learning
Fuzhou University
2011-2024
National University of Defense Technology
2021-2024
South China Agricultural University
2023
Ministry of Agriculture and Rural Affairs
2023
In recent years, the multiple traveling salesmen problem (MTSP or TSP) has received increasing research interest and one of its main applications is coordinated multirobot mission planning, such as cooperative search rescue tasks. However, it still challenging to solve MTSP with improved inference efficiency well solution quality in varying situations, e.g., different city positions, numbers cities, agents. this article, we propose an attention-based multiagent reinforcement learning (AMARL)...
Humans excel at reusing prior knowledge to address new challenges and developing skills while solving problems. This paradigm becomes increasingly popular in the development of autonomous agents, as it develops systems that can self-evolve response like human beings. However, previous methods suffer from limited training efficiency when expanding fail fully leverage facilitate task learning. In this paper, we propose Parametric Skill Expansion Composition (PSEC), a framework designed...
In recent years, cooperative Multi-Agent Deep Reinforcement Learning (MADRL) has received increasing research interest and been widely applied to computer games coordinated multi-robot systems, etc. However, it is still challenging realize high solution quality learning efficiency for MADRL under the conditions of incomplete noisy observations. To this end, paper proposes a multi-agent deep reinforcement approach with Grouped Cognitive featurE representatioN (GCEN), following paradigm...
Reinforcement learning from demonstration (RLfD) is considered to be a promising approach improve reinforcement (RL) by leveraging expert demonstrations as the additional decision‐making guidance. However, most existing RLfD methods only regard low‐level knowledge instances under certain task. Demonstrations are generally used either provide rewards or pretrain neural network‐based RL policy in supervised manner, usually resulting poor generalization capability and weak robustness...
Abstract Policy evaluation (PE) is a critical sub‐problem in reinforcement learning, which estimates the value function for given policy and can be used improvement. However, there still exist some limitations current PE methods, such as low sample efficiency local convergence, especially on complex tasks. In this study, novel algorithm called Least‐Squares Truncated Temporal‐Difference learning (LST 2 D) proposed. LST D, an adaptive truncation mechanism designed, effectively takes advantage...
Japanese encephalitis virus (JEV) is a typical mosquito-borne flavivirus that can cause central nervous system diseases in humans and animals. Host factors attempt to limit replication when the viruses invade host by using various strategies for replication. It essential clarify affect life cycle of JEV explore its underlying mechanism. Here, we found USP1-associated factor 1 (UAF1; also known as WD repeat-containing protein 48) modulated We propagation significantly increased UAF1-depleted...
Purpose – The purpose of this study is to evaluate the potential risks copyright infringement in digital library based on extension theory. Design/methodology/approach At first, analytic hierarchy process (AHP) used determine weights existing indicator system for early warning. Second, a model built theory library. Finally, real-world application presented show effectiveness and usefulness approach. Findings main findings paper are as follows: warning effective distinguishing degree library;...
In offline reinforcement learning, the challenge of out-of-distribution (OOD) is pronounced. To address this, existing methods often constrain learned policy through regularization. However, these suffer from issue unnecessary conservativeness, hampering improvement. This occurs due to indiscriminate use all actions behavior that generates dataset as constraints. The problem becomes particularly noticeable when quality suboptimal. Thus, we propose Adaptive Advantage-guided Policy...
In [8], Fang-Lan-Xiao proved a formula about Lusztig's induction and restriction functors which can induce Green's for the path algebra of quiver over finite field via trace map. this paper, we generalize their to that mixed semisimple perverse sheaves with an automorphism. By applying map, obtain any finite-dimensional hereditary field.
The present paper continues the work of [2]. For any symmetrizable generalized Cartan Matrix $C$ and corresponding quantum group $\mathbf{U}$, we consider associated quiver $Q$ with an admissible automorphism $a$. We construct category $\widetilde{\mathcal{Q}/\mathcal{N}}$ localization Lusztig sheaves for automorphism. Its Grothendieck gives a realization integrable highest weight $\mathbf{U}-$module $\Lambda_{\lambda}$, modulo traceless ones provide (signed) canonical basis...
Deep reinforcement learning (RL) typically requires a tremendous number of training samples, which are not practical in many applications. State abstraction and world models two promising approaches for improving sample efficiency deep RL. However, both state may degrade the performance. In this article, we propose an abstracted model-based policy (AMPL) algorithm, improves AMPL, novel method via multistep bisimulation is first developed to learn task-related latent spaces. Hence, original...