Adaptive Data Augmentation with Multi-armed Bandit: Sample-Efficient Embedding Calibration for Implicit Pattern Recognition

2026-02-22

AI Summary

This work addresses the challenge of fine-tuning pretrained foundation models for long-tailed implicit pattern recognition tasks, where scarce annotations and high computational costs hinder effective adaptation. To overcome these limitations, the authors propose ADAMAB, a framework that introduces a lightweight, embedder-agnostic calibrator on top of frozen embedding models and integrates an adaptive data augmentation strategy based on multi-armed bandits (MAB). By employing an improved upper confidence bound algorithm, ADAMAB mitigates gradient bias while providing theoretical convergence guarantees. Experimental results demonstrate that, under extreme few-shot settings with fewer than five initial samples per class, ADAMAB achieves up to a 40% absolute accuracy gain in multimodal scenarios, substantially reducing reliance on large-scale labeled datasets.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Adversarial Attacks & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Recognizing implicit visual and textual patterns is essential in many real-world applications of modern AI. However, tackling long-tail pattern recognition tasks remains challenging for current pre-trained foundation models such as LLMs and VLMs. While finetuning pre-trained models can improve accuracy in recognizing implicit patterns, it is usually infeasible due to a lack of training data and high computational overhead. In this paper, we propose ADAMAB, an efficient embedding calibration framework for few-shot pattern recognition. To maximally reduce the computational costs, ADAMAB trains embedder-agnostic light-weight calibrators on top of fixed embedding models without accessing their parameters. To mitigate the need for large-scale training data, we introduce an adaptive data augmentation strategy based on the Multi-Armed Bandit (MAB) mechanism. With a modified upper confidence bound algorithm, ADAMAB diminishes the gradient shifting and offers theoretically guaranteed convergence in few-shot training. Our multi-modal experiments justify the superior performance of ADAMAB, with up to 40% accuracy improvement when training with less than 5 initial data samples of each class.

Problem

• implicit pattern recognition
• long-tail recognition
• few-shot learning
• data scarcity
• foundation models

Innovation

• Adaptive Data Augmentation
• Multi-armed Bandit
• Embedding Calibration
• Few-shot Learning
• Sample Efficiency