Altair Yang YU - Homepage
Publications
2025
Conference Papers
Conference
2025
Asymmetric Pre-aligned Anchor Contrastive Enhanced Diffusion Hashing Model for Incomplete Multimodal Retrieval
ACM International Conference on Multimedia (ACM MM)
Abstract: Proposed a novel multi-modal diffusion hashing model for Incomplete Multi-modal Hashing Retrieval, featuring Asymmetric Pre-alignment strategy and Anchor Contrastive Enhanced Diffusion Hash mechanism.
2024
Conference Papers
Conference
2024
Unsupervised Multimodal Graph Contrastive Semantic Anchor Space Dynamic Knowledge Distillation Network for Cross-media Hash Retrieval
International Conference on Data Engineering (ICDE)
Abstract: Developed GASKN, a novel unsupervised multimodal graph contrastive semantic anchor space dynamic knowledge distillation network for cross-media hash retrieval.
Conference
2024
Knowledge Graph Enhanced Multimodal Transformer for Image-Text Retrieval
International Conference on Data Engineering (ICDE)
Abstract: Proposed a multimodal knowledge-enhanced multimodal transformer combining coarse-grained and fine-grained representation learning.
Conference
2024
Multimodal Knowledge Graph-guided Cross-Modal Graph Network for Image-text Retrieval
International Conference on Big Data and Smart Computing (BigComp)
Abstract: Constructed a novel multimodal knowledge graph-guided cross-modal graph network for fine-grained and coarse-grained image-text alignment.
Journal Papers
Journal
2024
Query Aware Cross-modal Dual Contrastive Learning Network for Multi-modal Video Moment Retrieval
Journal of Software
Abstract: Built a query-aware cross-modal contrastive learning network for multi-modal video moment retrieval (QACLN).
Journal
2024
Structures Aware Fine-Grained Contrastive Adversarial Hashing for Cross-Media Retrieval
IEEE Transactions on Knowledge and Data Engineering
Abstract: Established a cross-media contrastive adversarial hash network for cross-media hashing.
2022
Conference Papers
Conference
2022
Semantic Structure Enhanced Contrastive Adversarial Hash Network for Cross-media Representation Learning
30th ACM International Conference on Multimedia
Abstract: Developed SCAHN, a novel semantic structure enhanced contrastive adversarial hash network for cross-media representation learning.
Education
Doctor of Philosophy
2025 - Present
The Hong Kong Polytechnic University
Supervisor: Hongxia Yang
Research Interest: LLM Agents, Multimodal RAG, MLLM
Master's Degree
2022 - 2025
Beijing University of Posts and Telecommunications
Supervisor: Meiyu Liang
Research: Cross-modal Retrieval, Image-text Retrieval, Video Retrieval, Federated Learning
Bachelor's Degree
2018 - 2022
Hainan University
Major: Software Engineering
Work Experience
Professional Experience
Natural Language Algorithm Engineer
10/2023 - 11/2024
ByteDance (E-commerce on TikTok)
Backbone Construction Group for E-commerce Platform Governance
Key Responsibilities:
- Research and development of the backbone
- E-commerce platform content governance, model review and automatic rejection mechanism
- Enhanced performance of downstream tasks using pre-training techniques
- Developed novel pre-training approach combining LLM mask autoencoder and LLM next token prediction
- Achieved 7% increase in precision for downstream QA tasks
Natural Language Algorithm Engineer
04/2023 - 07/2023
AI Research Institute, New Oriental Education & Technology Group
Speech Language Group of Natural Language Processing Department
Key Responsibilities:
- Grammar error correction and intelligent question answering
- Optimized grammar error correction model
- Developed intelligent question-answering model
Academic Projects
Project Member
Cross-media Education Big Data Personalised Recommendation and Search System based on Deep Learning
National Natural Science Foundation of China
Project Member
Cross-modal Big Data Semantic Recognition and Search based on MindSpore
Chinese Association for Artificial Intelligence - Huawei MindSpore Academic Award Fund
Project Leader
Unified Semantic Representation Learning and Intelligent Search of Cross-Media Data Enhanced by Multi-modal Semantic Fusion
Graduate Innovation and Entrepreneurship Programme, Type A
Project Member
A Federated Learning Method for Quality and Efficiency Optimization of Distributed Associative Big Data
National Natural Science Foundation of China
Awards & Honors
2024
National Postgraduate Scholarship
November 2024
First Prize of Scholarship for Postgraduate Students
Beijing University of Posts and Telecommunications
November 2024
2023
Shenzhen Stock Exchange Enterprise Scholarship
November 2023
First Prize of Scholarship for Postgraduate Students
Beijing University of Posts and Telecommunications
November 2023
2022
First Prize of Scholarship for Postgraduate Students
Beijing University of Posts and Telecommunications
November 2022
Outstanding Graduate
Hainan University
September 2022
2021
First Prize in the Provincial Contest
Hainan in the 2020 China Collegiate Computing Contest - Group Programming Ladder Tournament
January 2021
2020
Merit Student
Hainan University
November 2020
First-class Comprehensive Scholarship
Hainan University
November 2020
Second Prize in Hainan Competition Area
National 3D Innovative Design Competition
October 2020
Third Prize in South China
2020 China Collegiate Computing Contest - WeChat Mini Programme Application Development Competition
Third Prize in the Intelligent Manufacturing Innovation and Creativity Contest
5th National Applied Talents Comprehensive Skills Competition, Hainan University
December 2019
2019
Special Comprehensive Scholarship
Hainan University
November 2019
Others
Research Areas
Cross-modal Retrieval, Image-text Retrieval, Video Retrieval
Multimodal Learning, Knowledge Distillation, Hash Learning
Federated Learning, Natural Language Processing
Large Language Models, RAG (Retrieval-Augmented Generation)
Software & Patents
Software Copyright
Quangou Hainan v1.0
Register No.: 2023SR0154035
Utility Model Patent
An Intelligent Huarong Dao Storage Cabinet
Register No.: 201922199084.7
Patent
A method and device for cross-modal hash retrieval based on multiple comparisons and dual-oppositional confrontation
Register No.: 202310700719.5
Languages & Interests
Languages
Mandarin (native)
English
Interests
Piano
Street Dance
Basketball
Badminton
Table Tennis