COLORA: Efficient Fine-Tuning for Convolutional Models with a Study Case on Optical Coherence Tomography Image Classification
Researchers have introduced COLORA (Convolutional Low-Rank Adaptation), a method that efficiently fine-tunes convolutional neural networks. COLORA reduces the number of trainable parameters by over 80% compared to full convolutional fine-tuning, while preserving model size and inference complexity. Experiments on medical image classification tasks demonstrate competitive performance with substantial parameter reduction.
Researchers have introduced COLORA (Convolutional Low-Rank Adaptation), a method that efficiently fine-tunes convolutional neural networks. COLORA reduces the number of trainable parameters by over 80% compared to full convolutional fine-tuning, while preserving model size and inference complexity. Experiments on medical image classification tasks demonstrate competitive performance with substantial parameter reduction.
---
Why it matters: This matters because it provides a more efficient way to fine-tune convolutional models for specific tasks, reducing the need for large amounts of training data and computational resources.
Source: https://arxiv.org/abs/2505.18315
This article was originally published at: https://arxiv.org/abs/2505.18315