QCQisen Chai

Qisen Chai

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Qisen Chai has a total of 11 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • PDSL: Propagation Dynamics Aware Framework for Source Localization

    by Yansong Wang, Qisen Chai, Longlong Lin, Tao Jia

    Source localization is a representative inverse inference task in information propagation, aiming to identify the source node or node set that triggers the propagation results based on the observed information. A primary challenge is quantifying the inherent uncertainty between observed outcomes and potential sources. Although deep generative models have partially mitigated this issue, most existing approaches primarily focus on uncertainty induced by network topology, attempting to learn a direct mapping from propagation outcomes to sources based on network structure, while overlooking the additional uncertainty stemming from the highly stochastic nature of the propagation process. To address this limitation, we propose a Propagation Dynamics aware framework for Source Localization (PDSL), a novel method that integrates a deep generative model with propagation dynamics to approximate the source distribution and explicitly mitigate uncertainty arising from diffusion stochasticity. Moreover, we employ Graph Neural Ordinary Differential Equations to model the continuous dynamics of diffusion processes without relying on a predefined diffusion mechanism. Additionally, a matching mechanism is designed to extract relevant data blocks that enhance source generation reliability. Comprehensive experiments on both synthetic and real-world diffusion datasets demonstrate the superior performance of the proposed framework across diverse application scenarios.


  • Open Access

    Learning to Compress Graphs via Dual Agents for Consistent Topological Robustness Evaluation

    by Qisen Chai, Yansong Wang, Junjie Huang, Tao Jia

    As graph-structured data grow increasingly large, evaluating their robustness under adversarial attacks becomes computationally expensive and difficult to scale. To address this challenge, we propose to compress graphs into compact representations that preserve both topological structure and robustness profile, enabling efficient and reliable evaluation. We propose Cutter, a dual-agent reinforcement learning framework composed of a Vital Detection Agent (VDA) and a Redundancy Detection Agent (RDA), which collaboratively identify structurally vital and redundant nodes for guided compression. Cutter incorporates three key strategies to enhance learning efficiency and compression quality: trajectory-level reward shaping to transform sparse trajectory returns into dense, policy-equivalent learning signals; prototype-based shaping to guide decisions using behavioral patterns from both high- and low-return trajectories; and cross-agent imitation to enable safer and more transferable exploration. Experiments on multiple real-world graphs demonstrate that Cutter generates compressed graphs that retain essential static topological properties and exhibit robustness degradation trends highly consistent with the original graphs under various attack scenarios, thereby significantly improving evaluation efficiency without compromising assessment fidelity.

  • Open Access

    Improved RRT* Path-Planning Algorithm Based on the Clothoid Curve for a Mobile Robot Under Kinematic Constraints

    by Kemeng Ran, Yujun Wang, Can Fang, Qisen Chai, Xingxiang Dong, Guohui Liu

    In this paper, we propose an algorithm based on the Rapidly-exploring Random Trees* (RRT*) algorithm for the path planning of mobile robots under kinematic constraints, aiming to generate efficient and smooth paths quickly. Compared to other algorithms, the main contributions of our proposed algorithm are as follows: First, we introduce a bidirectional expansion strategy that quickly identifies a direct path to the goal point in a short time. Second, a node reconnection strategy is used to eliminate unnecessary nodes, thereby reducing the path length and saving memory. Third, a path deformation strategy based on the Clothoid curve is devised to enhance obstacle avoidance and path-planning capability, ensuring collision-free paths that comply with the kinematic constraints of mobile robots. Simulation results demonstrate that our algorithm is simpler, more computationally efficient, expedites pathfinding, achieves higher success rates, and produces smoother paths compared to existing algorithms.


  • Open Access

    RJ-RRT: Improved RRT for Path Planning in Narrow Passages

    by Qisen Chai, Yujun Wang

    As a representative of sampling-based planning algorithms, rapidly exploring random tree (RRT), is extensively welcomed in solving robot path planning problems due to its wide application range and easy addition of nonholonomic constraints. However, it is still challenging for RRT to plan the path for configuration space with narrow passages. As a variant algorithm of RRT, rapid random discovery vine (RRV) gives a better solution, but when configuration space contains more obstacles instead of narrow passages, RRV performs slightly worse than RRT. In order to solve these problems, this paper re-examines the role of sampling points in RRT. Firstly, according to the state of the random tree expanding towards the current sampling point, a greedy sampling space reduction strategy is proposed, which decreases the redundant expansion of the random tree in space by dynamically changing the sampling space. Secondly, a new narrow passage judgment method is proposed according to the environment around of sampling point. After the narrow passage is identified, the narrow passage is explored by generating multiple subtrees inside the passage. The subtrees can be merged into the main tree that expands in a larger area by subsequent sampling. These improvements further enhance the value of sampling points. Compared with the existing RRT algorithms, the adaptability for different environments is improved, and the planning time and memory usage are saved.