KALAHash: Knowledge-Anchored Low-Resource Adaptation for Deep Hashing

2024-12-27

AI Summary

To address the severe performance degradation of deep hashing models under low-resource settings (i.e., extremely limited labeled samples) caused by distribution shift, this paper proposes a knowledge-enhanced, parameter-efficient hashing transfer framework. Methodologically, it introduces three core innovations: (1) a novel class-level textual knowledge anchoring mechanism that leverages semantic priors from pretrained language models to calibrate the hash space; (2) Class-Calibration LoRA (CLoRA), enabling class-aware, dynamic low-rank adaptation; and (3) Knowledge-Guided Discrete Optimization (KIDDO), which exploits text embeddings to guide discrete optimization of hash codes. Evaluated under cross-domain low-resource scenarios, the framework significantly improves retrieval accuracy and hash code discriminability while enhancing data utilization efficiency by 4×. It establishes a scalable, knowledge-driven paradigm for few-shot hashing learning.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Search and Optimization: Learning to Search
• Computer Vision: Learning & Optimization for CV

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Deep hashing has been widely used for large-scale approximate nearest neighbor search due to its storage and search efficiency. However, existing deep hashing methods predominantly rely on abundant training data, leaving the more challenging scenario of low-resource adaptation for deep hashing relatively underexplored. This setting involves adapting pre-trained models to downstream tasks with only an extremely small number of training samples available. Our preliminary benchmarks reveal that current methods suffer significant performance degradation due to the distribution shift caused by limited training samples. To address these challenges, we introduce Class-Calibration LoRA (CLoRA), a novel plug-and-play approach that dynamically constructs low-rank adaptation matrices by leveraging class-level textual knowledge embeddings. CLoRA effectively incorporates prior class knowledge as anchors, enabling parameter-efficient fine-tuning while maintaining the original data distribution. Furthermore, we propose Knowledge-Guided Discrete Optimization (KIDDO), a framework to utilize class knowledge to compensate for the scarcity of visual information and enhance the discriminability of hash codes. Extensive experiments demonstrate that our proposed method, Knowledge- Anchored Low-Resource Adaptation Hashing (KALAHash), significantly boosts retrieval performance and achieves a 4x data efficiency in low-resource scenarios.

Problem

• Deep Hashing
• Limited Resources
• Small Data Size

Innovation

• KALAHash
• CLoRA
• KIDDO