Bio-Inspired Fine-Tuning for Selective Transfer Learning in Image Classification

2026-01-16

IEEE Access

AI Summary

This work addresses the performance limitations of transfer learning in label-scarce image classification tasks caused by domain discrepancies between source and target datasets. To mitigate this issue, the authors propose BioTune, a novel method that, for the first time, integrates biologically inspired evolutionary optimization into the fine-tuning process. BioTune jointly and adaptively selects which layers to freeze while dynamically adjusting the learning rates of unfrozen layers, operating without manual intervention and remaining compatible with various CNN architectures. Extensive experiments across nine natural and medical image datasets demonstrate that BioTune consistently outperforms state-of-the-art approaches such as AutoRGN and LoRA, achieving superior performance across four mainstream CNN backbones and thereby validating its effectiveness and generalization capability.

Technology Category

• Machine Learning: Bio-inspired Learning
• Search and Optimization: Evolutionary Computation
• Computer Vision: Learning & Optimization for CV

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Deep learning has significantly advanced image analysis across diverse domains but often depends on large, annotated datasets for success. Transfer learning addresses this challenge by utilizing pre-trained models to tackle new tasks with limited labeled data. However, discrepancies between source and target domains can hinder effective transfer learning. We introduce BioTune, a novel adaptive fine-tuning technique utilizing evolutionary optimization. BioTune enhances transfer learning by optimally choosing which layers to freeze and adjusting learning rates for unfrozen layers. Through extensive evaluation on nine image classification datasets, spanning natural and specialized domains such as medical imaging, BioTune demonstrates superior accuracy and efficiency over state-of-the-art fine-tuning methods, including AutoRGN and LoRA, highlighting its adaptability to various data characteristics and distribution changes. Additionally, BioTune consistently achieves top performance across four different CNN architectures, underscoring its flexibility. Ablation studies provide valuable insights into the impact of BioTune’s key components on overall performance.

Problem

• transfer learning
• domain discrepancy
• image classification
• fine-tuning
• limited labeled data

Innovation

• Bio-inspired optimization
• adaptive fine-tuning
• selective transfer learning
• evolutionary algorithm
• layer-wise learning rate