Watch Wider and Think Deeper: Collaborative Cross-modal Chain-of-Thought for Complex Visual Reasoning
Published by
arXiv
About
Highly accomplished Ph.D. Candidate in Computer Science with extensive expertise in large-scale AI infrastructure, federated learning, and distributed optimization. Proven ability to develop cutting-edge algorithms for heterogeneous environments and privacy-preserving machine learning, demonstrated through significant contributions to over 20 peer-reviewed publications. Driven to build democratized and efficient AI systems, leveraging strong research and development skills to advance the state-of-the-art in LLM training and deployment.
Work
Alibaba Group
|Research Intern
Hangzhou, Zhejiang, China
→
Summary
As a Research Intern at Alibaba Group, Tao Shen collaborated with industry researchers to develop algorithms for large-scale AI infrastructure projects in heterogeneous distributed learning environments.
Highlights
Collaborated with leading industry researchers on large-scale AI infrastructure projects, advancing solutions for distributed computing environments.
Developed and optimized algorithms for heterogeneous distributed learning environments, improving efficiency and scalability of AI model training.
Zhejiang Lab
|Research Intern
Hangzhou, Zhejiang, China
→
Summary
As a Research Intern at Zhejiang Lab, Tao Shen focused on researching advanced machine learning algorithms and distributed systems, with a particular emphasis on federated learning and privacy-preserving techniques.
Highlights
Conducted in-depth research into advanced machine learning algorithms and distributed systems, contributing to novel theoretical and practical advancements.
Implemented and evaluated federated learning and privacy-preserving machine learning techniques, enhancing data security and model robustness for distributed AI applications.
Zhejiang University
|Research Assistant
Hangzhou, Zhejiang, China
→
Summary
As a Research Assistant at Zhejiang University, Tao Shen supported research on distributed optimization and control systems, developing numerical algorithms for critical power system analysis.
Highlights
Contributed to foundational research on distributed optimization and control systems, advancing understanding of complex networked systems.
Developed and refined numerical algorithms for power flow analysis and optimization, enhancing the efficiency and reliability of power system operations.
Education
Zhejiang University
Hangzhou, Zhejiang, China
→
Ph.D.
Computer Science
Courses
Federated Learning
Distributed Optimization
Trustworthy AI
Zhejiang University
Hangzhou, Zhejiang, China
→
M.S.
Control Science and Engineering
Courses
Power Systems
Numerical Methods
Optimization
China University of Petroleum
Qingdao, Shandong, China
→
B.S.
Automation
Grade: Top-tier Class
Courses
Neural Network
Control Systems
Awards
Graduate Academic Excellence Scholarship
Awarded By
Zhejiang University
Recognized for outstanding academic performance and research contributions at the graduate level.
National Innovation Program Fellowship
Awarded By
China University of Petroleum
Awarded for exceptional potential in innovation and research during undergraduate studies.
Research Innovation Scholarship
Awarded By
College of Control Science & Engineering
Received for demonstrating significant aptitude and contributions to research within the college.
Publications
Remedy: Recipe Merging Dynamics in Large Vision-Language Models
Published by
Proc. ICLR
Improving Model Fusion by Training-time Neuron Alignment with Fixed Neuron Anchors
Published by
IEEE Trans. Pattern Anal. Mach. Intell.
Knowledge-empowered, Collaborative, and Co-evolving AI Models: The Post-LLM Roadmap
Published by
Engineering
Each Rank Could Be an Expert: Single-ranked Mixture of Experts LoRA for Multi-task Learning
Published by
arXiv
FedCFA: Alleviating Simpson's Paradox in Model Aggregation with Counterfactual Federated Learning
Published by
Proc. AAAI
FlowerTune: A Cross-Domain Benchmark for Federated Fine-Tuning of Large Language Models
Published by
arXiv
Fedeve: On Bridging the Client Drift and Period Drift for Cross-device Federated Learning
Published by
arXiv
Will LLMs Scaling Hit the Wall? Breaking Barriers via Distributed Resources on Massive Edge Devices
Published by
arXiv
FedGuCci: Making Local Models More Connected in Landscape for Federated Learning
Published by
Proc. ACM SIGKDD Conf. Knowl. Discovery Data Mining
Retrieval-augmented Mixture of LoRA Experts for Uploadable Machine Learning
Published by
arXiv
Training-time Neuron Alignment through Permutation Subspace for Improving Linear Mode Connectivity and Model Fusion
Published by
arXiv
Improving Group Connectivity for Generalization of Federated Deep Learning
Published by
Proc. Int. Workshop Federated Foundation Models
An Adaptive Aggregation Method for Federated Learning via Meta Controller
Published by
Proc. ACM Int. Conf. Multimedia Asia
Merging LoRAs Like Playing Lego: Pushing the Modularity of LoRA to Extremes through Rank-wise Clustering
Published by
arXiv
Model Tailor: Mitigating Catastrophic Forgetting in Multi-modal Large Language Models
Published by
arXiv
Deconfounded Hierarchical Multi-granularity Classification
Published by
Comput. Vis. Image Underst.
Federated Unsupervised Representation Learning
Published by
Front. Inf. Technol. Electron. Eng.
Duet: A Tuning-free Device-cloud Collaborative Parameters Generation Framework for Efficient Device Model Generalization
Published by
Proc. ACM Web Conf.
Edge-cloud Polarization and Collaboration: A Comprehensive Survey for AI
Published by
IEEE Trans. Knowl. Data Eng.
FedDTG: Federated Data-Free Knowledge Distillation via Three-Player Generative Adversarial Networks
Published by
arXiv
Federated Graph Learning-A Position Paper
Published by
arXiv
Federated Mutual Learning
Published by
arXiv
A Graph-based Power Flow Method for Balanced Distribution Systems
Published by
Energies
Languages
English
Chinese (Mandarin)
Skills
Programming Languages
Python, LaTeX/Beamer, MATLAB, C/C++, SQL.
ML/DL Frameworks
PyTorch, Flower (FL framework), PEFT (LoRA).
Application Development
Web, Browser Extensions, iOS/macOS Apps.
Development Tools
GitHub, Hugging Face, Docker.
Visualization
Matplotlib, Plotly, Manim.