About Me
Wang Jie (王杰)
I completed my undergraduate studies in the Department of Computer Science and Technology at Tongji University and am currently working as a research assistant at the MAPLE Lab at Westlake University. My research interests include multimodal large models, safety supervision mechanisms and simulations for autonomous driving, as well as virtual reality. Currently, I am primarily focused on exploring the applications of large models in fine-grained emotional computation and multimodal recommendation systems.
During my undergraduate studies, I led a team to establish two startups aimed at developing mental health applications based on extended reality technologies and large language models. Our team has successfully developed two applications, which are actively being improved and tested. The project has received support from two startup funds and has won over ten awards in various innovation and entrepreneurship competitions both domestically and internationally.
If you are interested in my research directions or startup projects, or are looking for any form of collaboration, please feel free to email me at 2054310@tongji.edu.cn.
Educations
.png)
Tongji University
Bachalor Degree(2020 - 2024)
SRIAS / Tongji
Research Assistant(2022 - 2023)
Westlake University
Research Assistant(2024 - )
......
Publicatons
SentiXRL: An advanced large language Model Framework for
Multilingual Fine-Grained Emotion Classification in Complex Text
Environment (COLING 2025 under
review)
Jie Wang, Yichen Wang, Zhilin Zhang, Kaidi Wang,
Zhiyang Chen*
With strong expressive
capabilities in Large Language Models(LLMs), generative models
effectively capture sentiment structures and deep semantics, however,
challenges remain in fine-grained sentiment classification across
multi-lingual and complex contexts. To address this, we propose the
Sentiment Cross-Lingual Recognition and Logic Framework (SentiXRL),
which incorporates two modules, an emotion retrieval enhancement module
to improve sentiment classification accuracy in complex contexts through
historical dialogue and logical reasoning,and a self-circulating
analysis negotiation mechanism (SANM)to facilitates autonomous
decision-making within a single model for classification
tasks.
VTD: Visual and Tactile Database for Driver State and
Behavior Perception (IEEE RAL under
review)
Jie Wang, Mobing Cai, Zhongpan Zhu*, Member, IEEE,
Hongjun Ding, Jiwei Yi and Aimin Du
In the domain of autonomous vehicles,
the human vehicle co-pilot system has garnered significant research
attention. To address the subjective uncertainties in driver state and
interaction behaviors, which are pivotal to the safety of
Human-in-the-loop co-driving systems, we introduce a novel
visual-tactile perception method. Utilizing a driving simulation
platform, a comprehensive dataset has been developed that encompasses
multi-modal data under fatigue and distraction conditions. The
experimental setup integrates driving simulation with signal
acquisition, yielding 600 minutes of fatigue detection data from 15
subjects and 102 takeover experiments with 17 drivers. The dataset,
synchronized across modalities, serves as a robust resource for
advancing cross-modal driver behavior perception algorithms.
Perturbation Ontology based Graph Attention Networks (ICASSP 2025 under review)
Yichen Wang, Jie Wang, Fulin Wang, Xiang Li, Hao Yin,
Bhiksha Raj*
In recent years, graph representation
learning has undergone a paradigm shift, driven by the emergence and
proliferation of graph neural networks (GNNs) and their heterogeneous
counterparts. Heterogeneous GNNs have shown remarkable success in
extracting low-dimensional embeddings from complex graphs that encompass
diverse entity types and relationships. While meta-path-based techniques
have long been recognized for their ability to capture semantic
affinities among nodes, their dependence on manual specification poses a
significant limitation. In this paper, we challenge the current paradigm
by introducing ontology as a fundamental semantic primitive within
complex graphs. Our goal is to integrate the strengths of both
matrixcentric and meta-path-based approaches into a unified framework.
We propose perturbation Ontology-based Graph Attention Networks (POGAT),
a novel methodology that combines ontology subgraphs with an advanced
self-supervised learning paradigm to achieve a deep contextual
understanding. Through extensive empirical evaluations, we demonstrate
that POGAT significantly outperforms state-of-the-art baselines,
achieving a groundbreaking improvement of up to 14.46% in F1-score for
the critical task of link prediction and 15.76% in Micro-F1 for the
critical task of node classification.
Perturbation Ontology based Graph Attention Networks (AAAI Student Program 2025 under review)
Yichen Wang, Jie Wang, Fulin Wang, Xiang Li, Hao Yin,
Bhiksha Raj*
In recent years, graph representation
learning has undergone a paradigm shift, driven by the emergence and
proliferation of graph neural networks (GNNs) and their heterogeneous
counterparts. Heterogeneous GNNs have shown remarkable success in
extracting low-dimensional embeddings from complex graphs that encompass
diverse entity types and relationships. While meta-path-based techniques
have long been recognized for their ability to capture semantic
affinities among nodes, their dependence on manual specification poses a
significant limitation. In this paper, we challenge the current paradigm
by introducing ontology as a fundamental semantic primitive within
complex graphs. Our goal is to integrate the strengths of both
matrixcentric and meta-path-based approaches into a unified framework.
We propose perturbation Ontology-based Graph Attention Networks (POGAT),
a novel methodology that combines ontology subgraphs with an advanced
self-supervised learning paradigm to achieve a deep contextual
understanding. Through extensive empirical evaluations, we demonstrate
that POGAT significantly outperforms state-of-the-art baselines,
achieving a groundbreaking improvement of up to 14.46% in F1-score for
the critical task of link prediction and 15.76% in Micro-F1 for the
critical task of node classification.
Efficient Bilinear Attention-based Fusion for Medical Visual
Question Answering (BIBM 2025 final
decision)
Zhilin Zhang, Jie Wang, Ruiyang Qin, Ruiqi Zhu and
Xiaoliang Gong*
Medical Visual Question Answering
(MedVQA) is a crucial task that combines computer vision and natural
language processing to assist in clinical decision-making by answering
questions based on medical images. In this paper, we propose the
Intra-modal and Cross-modal Bilinear Attention Network (ICBAN), a novel
model that integrates multi-scale feature extraction, self-attention
mechanisms, and bilinear attention networks to effectively fuse visual
and textual features. Our approach addresses the challenges of MedVQA by
balancing accuracy and computational efficiency. Experimental results
demonstrate that IC-BAN can achieve superior performance compared to
both traditional BAN and Transformer-based fusion methods. This work
highlights the potential of advanced bilinear attention mechanisms in
MedVQA.
Enhancing Autonomous
Driving Safety: A Robust Stacking Ensemble Model for Traffic Sign
Detection and Recognition (Sustainability
2025)
Yichen Wang, Jie Wang, Qianjin Wang
Accurate detection and classification of traffic
signs play a vital role in ensuring driver safety and supporting
advancements in autonomous driving technology. This paper introduces
a novel approach for traffic sign detection and recognition by
integrating the Faster RCNN and YOLOX-Tiny models using a stacking
ensemble technique. The innovative ensemble methodology creatively
merges the strengths of both models, surpassing the limitations of
individual algorithms and achieving superior performance in
challenging real-world scenarios. Our results show improved accuracy
and efficiency in recognizing traffic signs in various real-world
scenarios, including distant, close, complex, moderate, and simple
settings, achieving a 4.78% increase in mean Average Precision (mAP)
compared to Faster RCNN and improving Frames Per Second (FPS) by
8.1% and mAP by 6.18% compared to YOLOX-Tiny.
Intellectual Property
A safety officer workload estimation system and estimation method
for 5G remote driving, China [P]. 202211634308.2, 2022.
Iland - A psychological healing program based on virtual display
technology, China. Software Copyright Registration No. 2023SR0563014, 2023.
Sandbox Online - virtual sandbox system, China. Software Copyright
Registration No. 2023SR0676789, 2023.
Honors and Awards
2023 Global Competition on Design For Future
Education
Best Design Award
2023 Chinese Artifical Intelligence Application
Scenario Innovation Challenge
Merit Award
2023 China-U.S. Young Maker Competition Shanghai
Division
Third Prize
2023 The 16th National Student Software
Innovation
Semifinalist (Top 60)
2023 TCCI AI for Brain Science Competition
First Prize
2023 The 8th Huichuang Youth Shanghai Student
Projects Competiton
First Prize
2022 IFLYTEK AI Development Competition
Merit Award (4/102)
2023 The 2nd China-Africa Youth Innovation and
Entrepreneurship Competition
Merit Award
Internship & Summer School
Nokia Shanghai Bell 2022 Summer school(Nokia
NILC-2022) Best Progress Award Jul. 2022 - Aug. 2022 [Link]
2023 Baidu AI talent development Campus
Graduate Excellence Award Jan. 2022 - Nov.
2022 [Link]
2024 Westlake AI Open Courses
Best Project Runner-up Award Jul. 2024 -
Aug. 2024 [Link]
Entrepreneurship
Shanghai Besen Technology Co.,Ltd. (BR No. 310110001229712,Certificate
of Organization Code:MAC5RUKW-1)
Co-Founder,CEO
● Incubated enterprises in National Science and
Technology Park and Venture Valley of Tongji University, backed by investment agency
funds (Shanghai Tongji Technology Incubator Co.).
● Developed a well-rounded, multi-platform application using VR/AR technology and deep learning to combine traditional psychological healing techniques and reduce the reliance on psychological consultants.
● Established and led a startup team, expanding the team to nearly 50 people.
● Developed a well-rounded, multi-platform application using VR/AR technology and deep learning to combine traditional psychological healing techniques and reduce the reliance on psychological consultants.
● Established and led a startup team, expanding the team to nearly 50 people.
Project Introduction
Considering the increasingly deteriorating mental health of the younger population and the growing risks of psychological issues, coupled with the shortage of mental health professionals and the general public's lack of awareness of mental health, there is a noticeable gap in applications addressing mental healing and stress relief. While AR technology has advanced significantly, particularly in mobile applications, it is predominantly used in B2B settings, mainly for product promotion, with limited progress in other fields. Taking into account the current market landscape and technological considerations, we have developed a multi-platform XR interactive application focused on mental healing and stress relief, providing an innovative solution for psychological well-being.
