Tongji University — Publications

Enhancing Environmental Robustness in Few-shot Learning via Conditional Representation Learning

Article

Few-shot learning (FSL) has recently been extensively utilized to overcome the scarcity of training data in domain-specific visual recognition. In real-world scenarios, environmental factors such as complex backgrounds, varying lighting conditions, long-distance shooting, and moving targets often cause test images to exhibit numerous incomplete targets or noise disruptions. However, current research on evaluation datasets and methodologies has largely ignored the concept of "environmental robustness", which refers to maintaining consistent performance in complex and diverse physical environments. This neglect has led to a notable decline in the performance of FSL models during practical testing compared to their training performance. To bridge this gap, we introduce a new real-world multi-domain few-shot learning (RD-FSL) benchmark, which includes four domains and six evaluation datasets. The test images in this benchmark feature various challenging elements, such as camouflaged objects, small targets, and blurriness. Our evaluation experiments reveal that existing methods struggle to utilize training images effectively to generate accurate feature representations for challenging test images. To address this problem, we propose a novel conditional representation learning network (CRLNet) that integrates the interactions between training and testing images as conditional information in their respective representation processes. The main goal is to reduce intra-class variance or enhance inter-class variance at the feature representation level. Finally, comparative experiments reveal that CRLNet surpasses the current state-of-the-art methods, achieving performance improvements ranging from 6.83% to 16.98% across diverse settings and backbones. The source code and dataset are available at https://github.com/guoqianyu-alberta/Conditional-Representation-Learning.

Qianyu Guo, Jingrong Wu, Tianxing Wu et al. 2025

Publications

Cognitive Bias and Reassignment: Who Can Contribute High Quality LLM Data

Risk factors of heart failure for patients classification with extreme learning machine

SemSearch'11

Towards better understanding and utilizing relations in DBpedia

U-NIAH: Unified RAG and LLM Evaluation for Long Context Needle-In-A-Haystack

BSBM+: Extending BSBM to Evaluate Annotated RDF Features on Graph Databases

Enhancing Environmental Robustness in Few-shot Learning via Conditional Representation Learning

Meta-CoT-A*-MCTS: Search for Stronger User Preference Alignment in Agent4Rec

A semantic approach to recommending text advertisements for images

Proceedings of the 3rd International Semantic Search Workshop

COKG-QA: Multi-hop Question Answering over COVID-19 Knowledge Graphs

Knowledge graph construction from multiple online encyclopedias

Software.zhishi.schema: A Software Programming Taxonomy Derived from Stackoverflow.

Knowledge-Enhanced Medical Visual Question Answering: A Survey (Invited Talk Summary)

Rewarding What Matters: Step-by-Step Reinforcement Learning for Task-Oriented Dialogue

Enhancing Environmental Robustness in Few-shot Learning via Conditional Representation Learning

Building Large-Scale Knowledge Base for Relations from Text

The Semantic Web and Web Science

Efficient query expansion for advertisement search

The allelopathy and allelopathic mechanism of phenolic acids on toxic Microcystis aeruginosa