Author name cluster

Jin Jiang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

6 papers

1 author row

EAAI Journal 2025 Journal Article

A cascaded strategy-based hierarchical reinforcement learning algorithm for lane change decision-making

Heng Du
Dingfa Lin
Xiaolong Zhang
Lingtao Wei
Shizhao Zhou
Xuanhao Cheng
Luxin Zhang
Jin Jiang

Hierarchical reinforcement learning (HRL) has demonstrated considerable promise in addressing complex driving tasks. However, existing HRL-based autonomous driving decision systems face challenges such as inefficient convergence, lack of interdependence among driving maneuver strategies (including throttle/brake control and steering adjustments), and inadequate risk assessment mechanisms, all of which impede the safety and stability of lane-changing decisions. This study proposes a novel HRL framework for continuous lane-changing decision planning. This framework establishes cascaded relationships between driving maneuvers strategies and integrates a comprehensive risk assessment mechanism to address these challenges. Initially, a hierarchical decision model is developed, where the high-level determines the lane-changing intent, while the low-level manages continuous and precise maneuvers. Subsequently, by integrating a Bayesian network, the cascading between throttle/brake openings and steering angles is achieved, optimizing the system's joint strategy distribution. Furthermore, a comprehensive risk assessment mechanism that evaluates the cooperation level of drivers and the severity of potential collisions is designed to encourage agents to adopt strategies that minimize risk. The effectiveness of the proposed decision-making framework has been validated through comparative experiments in mixed traffic scenarios simulated within the Car Learning to Act (CARLA) environment and corroborated with human driving data from the Next Generation Simulation (NGSIM) database.

Details DOI

AAAI Conference 2025 Conference Paper

S^3cMath: Spontaneous Step-Level Self-Correction Makes Large Language Models Better Mathematical Reasoners

Yuchen Yan
Jin Jiang
Yang Liu
Yixin Cao
Xin Xu
Mengdi Zhang
Xunliang Cai
Jian Shao

Self-correction is a novel method that can stimulate the potential reasoning abilities of large language models (LLMs). It involves detecting and correcting errors during the inference process when LLMs solve reasoning problems. However, recent works do not regard self-correction as a spontaneous and intrinsic capability of LLMs. Instead, such correction is achieved through post-hoc generation, external knowledge introduction, multi-model collaboration, and similar techniques. In this paper, we propose a series of mathematical LLMs called S^3cMath, which are able to perform Spontaneous Step-level Self-correction for Mathematical reasoning. This capability helps LLMs to recognize whether their ongoing inference tends to contain errors and simultaneously correct these errors to produce a more reliable response. We proposed a method, which employs a step-level sampling approach to construct step-wise self-correction data for achieving such ability. Additionally, we implement a training strategy that uses above constructed data to equip LLMs with spontaneous step-level self-correction capacities. Our data and methods have been demonstrated to be effective across various foundation LLMs, consistently showing significant progress in evaluations on GSM8K, MATH, and other mathematical benchmarks. To the best of our knowledge, we are the first to introduce the spontaneous step-level self-correction ability of LLMs in mathematical reasoning.

PDF Details DOI

YNIMG Journal 2025 Journal Article

Sleep indicators and staging: A functional near-infrared spectroscopy study in healthy young adults

Yong Cao
Xingwei An
Wenxiao Zhong
Jin Jiang
Hongzuo Chu
Xuejun Jiao
Xiaoping Chen
Yufeng Ke

Functional near-infrared spectroscopy(fNIRS)-based sleep staging has attracted considerable interest due to its portability and limited interference with sleep. However, few studies have systematically examined sleep indicators or formulated sleep staging models based on fNIRS features labelled by polysomnography(PSG). This study aimed to address these shortcomings and promote the application of fNIRS in sleep monitoring. 37 volunteers participated in our experiment, with 6-channel prefrontal fNIRS data and standard PSG data collected simultaneously. Sleep indicators were extracted from time-domain, frequency-domain, and entropy perspectives. Sleep staging was developed based on these indicators using human-scored PSG as reference. Our findings indicated deeper sleep was correlated with a decrease in amplitude of time-domain features, while entropy features showed a contrasting trend. The fNIRS-based sleep staging achieved a Cohen's kappa(κ) of 0.76±0.12, 0.72±0.09, 0.71±0.07, with accuracies of 94.2 ± 2.4 %, 87.8 ± 3.2 %, and 82.2 ± 4.1 %, for 2-class(Wake/Sleep), 3-class(Wake/NREM/REM), 4-class (Wake/N1+N2/N3/REM) classifications, respectively. Sleep statistics derived from fNIRS closely aligned with those from PSG, with differences in sleep onset latency, wake after sleep onset, total wake/sleep time within 5 min and sleep efficiency below 3 %. The substantial agreement in both detailed (epoch-by-epoch) and comprehensive (total) sleep statistics with PSG suggests fNIRS is a reliable tool for long-term sleep monitoring in everyday settings.

Details DOI

NeurIPS Conference 2025 Conference Paper

Uni-MuMER: Unified Multi-Task Fine-Tuning of Vision-Language Model for Handwritten Mathematical Expression Recognition

Yu Li
Jin Jiang
Jianhua Zhu
Shuai Peng
Yuxuan Zhou
Liangcai Gao

Handwritten Mathematical Expression Recognition (HMER) remains a persistent challenge in Optical Character Recognition (OCR) due to the inherent freedom of symbol layouts and variability in handwriting styles. Prior methods have faced performance bottlenecks by proposing isolated architectural modifications, making them difficult to integrate coherently into a unified framework. Meanwhile, recent advances in pretrained vision-language models (VLMs) have demonstrated strong cross-task generalization, offering a promising foundation for developing unified solutions. In this paper, we introduce Uni-MuMER, which fully fine-tunes a VLM for the HMER task without modifying its architecture, effectively injecting domain-specific knowledge into a generalist framework. Our method integrates three data-driven tasks: Tree-Aware Chain-of-Thought (Tree-CoT) for structured spatial reasoning, Error-Driven Learning (EDL) for reducing confusion among visually similar characters, and Symbol Counting (SC) for improving recognition consistency in long expressions. Experiments on the CROHME and HME100K datasets show that Uni-MuMER achieves super state-of-the-art performance, outperforming the best lightweight specialized model SSAN by 16. 31\% and the top-performing VLM Gemini2. 5-flash by 24. 42\% under zero-shot setting. Our datasets, models, and code are open-sourced at: https: //github. com/BFlameSwift/Uni-MuMER

PDF Details

AAAI Conference 2024 Conference Paper

History Matters: Temporal Knowledge Editing in Large Language Model

Xunjian Yin
Jin Jiang
Liming Yang
Xiaojun Wan

The imperative task of revising or updating the knowledge stored within large language models arises from two distinct sources: intrinsic errors inherent in the model which should be corrected and outdated knowledge due to external shifts in the real world which should be updated. Prevailing efforts in model editing conflate these two distinct categories of edits arising from distinct reasons and directly modify the original knowledge in models into new knowledge. However, we argue that preserving the model's original knowledge remains pertinent. Specifically, if a model's knowledge becomes outdated due to evolving worldly dynamics, it should retain recollection of the historical knowledge while integrating the newfound knowledge. In this work, we introduce the task of Temporal Knowledge Editing (TKE) and establish a benchmark AToKe (Assessment of TempOral Knowledge Editing) to evaluate current model editing methods. We find that while existing model editing methods are effective at making models remember new knowledge, the edited model catastrophically forgets historical knowledge. To address this gap, we propose a simple and general framework termed Multi-Editing with Time Objective (METO) for enhancing existing editing models, which edits both historical and new knowledge concurrently and optimizes the model's prediction for the time of each fact. Our assessments demonstrate that while AToKe is still difficult, METO maintains the effectiveness of learning new knowledge and meanwhile substantially improves the performance of edited models on utilizing historical knowledge.

PDF Details DOI

EAAI Journal 2020 Journal Article

Enhanced fault diagnosis method using conditional Gaussian network for dynamic processes

Chuyue Lou
Xiangshun Li
M. Amine Atoui
Jin Jiang

Applying fault detection and diagnosis (FDD) technology to the process industry can help to detect faults in time and minimize their impact. The purpose of this study is to propose an enhanced fault diagnosis method under a Conditional Gaussian Network(CGN) efficient and suitable for dynamic processes fault monitoring. The key paths are as follows: first, a time series model is established for the process data and decomposed into time-dependent components and time-independent components; second, time-dependent components are discarded and time-independent components void of auto-correlation are considered instead of the original data to learn the CGN model. A numerical simulation case is used to illustrate the interest of our proposal. The effectiveness of the proposed method is further verified and compared on the Tennessee Eastman Process (TEP). The obtained results show that our method has high and better accuracies regarding the diagnosis of known and unknown faults in dynamic processes.

Details DOI