Provably Improving Generalization of Few-Shot Models with Synthetic Data
Authors:
- Lan-Cuong NguyenLNLan-Cuong Nguyen
- Quan Nguyen-TriQNQuan Nguyen-Tri
- Bang Tran KhanhBKBang Tran Khanh
- Dung D. LeDLDung D. Le
- Long Tran-ThanhLTLong Tran-Thanh
- Khoat ThanKTKhoat Than
Preprint
Abstract
Few-shot image classification remains challenging due to the scarcity of labeled training examples. Augmenting them with synthetic data has emerged as a promising way to alleviate this issue, but models trained on synthetic samples often face performance degradation due to the inherent gap between real and synthetic distributions. To address this limitation, we develop a theoretical framework that quantifies the impact of such distribution discrepancies on supervised learning, specifically in the context of image classification. More importantly, our framework suggests practical ways to generate good synthetic samples and to train a predictor with high generalization ability. Building upon this framework, we propose a novel theoretical-based algorithm that integrates prototype learning to optimize both data partitioning and model training, effectively bridging the gap between real few-shot data and synthetic data. Extensive experiments results show that our approach demonstrates superior performance compared to state-of-the-art methods, outperforming them across multiple datasets.Keywords
All Related Versions
This publication has 7 references indexed in Scilit:
- Diversified In-domain Synthesis with Efficient Fine-Tuning for Few-Shot Classification2026
- The Faiss LibraryIEEE Transactions on Big Data, 2025
- Gentle local robustness implies generalizationMachine Learning, 2025
- Large Language Models are Good Prompt Learners for Low-Shot Image Classification2024
- Prototype-based Dataset Comparison2023
- EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover ClassificationIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019
- Robust Classification with Convolutional Prototype Learning2018