- Reinforcement Learning in Robotics
- Adversarial Robustness in Machine Learning
- Autonomous Vehicle Technology and Safety
- Anomaly Detection Techniques and Applications
- Peer-to-Peer Network Technologies
- Advanced Data Storage Technologies
- Generative Adversarial Networks and Image Synthesis
- Food Supply Chain Traceability
- Accounting and Organizational Management
- Computer Graphics and Visualization Techniques
- Risk Management in Financial Firms
- Supply Chain Resilience and Risk Management
- Smart Grid Security and Resilience
- Economic and Technological Developments in Russia
- Face recognition and analysis
- Caching and Content Delivery
- Cellular Automata and Applications
- Traffic control and management
- Machine Learning and Algorithms
- Adaptive Dynamic Programming Control
- Blockchain Technology Applications and Security
- Game Theory and Applications
- Complex Network Analysis Techniques
- Physical Education and Training Studies
- Domain Adaptation and Few-Shot Learning
Harbin Institute of Technology
2024-2025
Tongji University
2021-2024
Capital Medical University
2024
Beijing Tongren Hospital
2024
Tsinghua University
2024
Beijing Forestry University
2024
University of Glasgow
2022
University of Sheffield
2016
Mineral Products Association
2013
McGill University
2008
Metaverse is an artificial virtual world mapped from and interacting with the real world. In metaverse, digital entities coexist their physical counterparts. Powered by deep learning, metaverse inevitably becoming more intelligent in interactions between reality virtuality. However, it confronted a nontrivial problem known as sim2real transfer when learning techniques try to bridge gap simulations. this article, we use multiagent reinforcement (MARL) implement collective intelligence for...
A control system of multiple unmanned aerial vehicles (multi-UAV) is generally very complex when they complete a task in closely-cooperative manner, e.g. two UAVs cooperatively transport package goods. Multi-agent reinforcement learning (MARL) offers promising solution for such control. However, MARL heavily relies on trial-and-error explorations, facing big challenge gathering real-world training data. Simulation environments are commonly used to overcome this challenge, i.e., policy...
The most important asset of a Massively Multiplayer Online Game is its world state, as it represents the combined efforts and progress all participants. Thus, extremely that this state not lost in case server failures. Survival typically achieved by making persistent, e.g., storing relational database. main challenge approach to track large volume modifications applied real time. This paper compares variety strategies persist changes game world. While critical events must be written...
Unmanned Aerial Vehicles (UAVs) have extensive applications such as logistics transportation and aerial photography. However, UAVs are sensitive to winds. Traditional control methods, proportional- integral-derivative controllers, generally fail work well when the strength direction of winds changing frequently. In this deep reinforcement learning algorithms combined with a domain randomization method learn robust wind-resistant hovering policies. A novel reward function is designed guide...
Zero-shot learning (ZSL) enables models to recognize categories not encountered during training, which is crucial for with limited data. Existing methods overlook efficient temporal modeling in multimodal This paper proposes a Temporal–Semantic Aligning and Reasoning Transformer (TSART) spatio-temporal modeling. TSART uses the pre-trained SeLaVi network extract audio visual features explores semantic information of these modalities through encoders. It incorporates reasoning module enhance...
Recent advancements in connected automated vehicles (CAVs) and reinforcement learning (RL) hold significant promise for enhancing intelligent traffic control systems. This paper conducts a systematic review of studies on RL-based urban at signalised intersections, highlighting the impact CAVs performance improvement. We first fundamental concepts RL algorithms, establishing foundational understanding subsequent methods. then recent progress signal using CV/CAV trajectory data, CAV planning,...
At present, the global market economy is in stage of rapid development, and competition among enterprises becoming much fiercer. The operation management need to adapt changes market, which requires constantly explore external markets, strengthen management, reduce product costs, improve core competitiveness. change accounting a kind conscious spontaneous behavior internal environment enterprises. aim this article critically discuss construction control system its influence with...
In areas such as finance, engineering, and science, we often face situations that change quickly unpredictably. These are tough to handle require special tools methods capable of understanding predicting what might happen next. Stochastic Differential Equations (SDEs) renowned for modeling analyzing real-world dynamical systems. However, obtaining the parameters, boundary conditions, closed-form solutions SDEs can be challenging. this paper, will discuss application Kalman filtering theory...
This paper considers discounted infinite horizon mean field games by extending the probabilistic weak formulation of game as introduced Carmona and Lacker (2015). Under similar assumptions in finite game, we prove existence uniqueness solutions for extended game. The key idea is to construct local versions previously considered stable topologies. Further, analyze how sequences approximate one. a weakened Lasry-Lions monotonicity condition, can quantify convergence rate one using novel...
Imitation learning is a promising approach to extract knowledge from human’s demonstrations. In traditional imitation methods like behavioral cloning, human demonstration transitions were uniformly sampled replay buffer regardless of their different values. addition, agent trained by this method limited solve specific task. paper, we extend cloning with prioritized sampling technique. To make our more general, introduce an additional element goal which the difference between last state...
Simulation based decision making tools, such as simulation cloning, "what-if" analysis, and etc., has being a beneficial way to analyzing of multiple alternative scenarios, however, there is no guarantee that simulator could obtain feasible scenario meeting the flood mitigation requirements, let alone an optimal one.Motivated by J. R. Marden his colleague's work "cooperative control potential games", novel technique, selection game, was proposed in this paper solve optimization problem,...
Diffusion models have achieved remarkable image generation quality surpassing previous generative models. However, a notable limitation of diffusion models, in comparison to GANs, is their difficulty smoothly interpolating between two samples, due highly unstructured latent space. Such smooth interpolation intriguing as it naturally serves solution for the morphing task with many applications. In this work, we present DiffMorpher, first approach enabling and natural using Our key idea...
The advent of the Internet has changed modern life, creating new opportunities in areas such as work, study and leisure. However, a result this technology, many problems have also emerged at same time. purpose paper is to examine an important issue related within financial industry methods solve it. This shows that credit risk one biggest risks internet finance virtual trading platform. increased default caused by information asymmetry brings two main consequences, namely inability recover...