CVAILGIVDec 29, 2023

FerKD: Surgical Label Adaptation for Efficient Distillation

arXiv:2312.17473v17 citationsh-index: 4Has CodeICCV
AI Analysis

This work addresses efficiency and accuracy issues in knowledge distillation for computer vision researchers, offering a novel method that reduces heuristics and hyperparameters compared to prior solutions.

The paper tackles the problem of inefficient knowledge distillation by proposing FerKD, a framework that uses partial soft-hard label adaptation and region-calibration to improve reliability in teacher predictions, achieving 81.2% accuracy on ImageNet-1K with ResNet-50 and up to 89.9% with ViT-G14.

We present FerKD, a novel efficient knowledge distillation framework that incorporates partial soft-hard label adaptation coupled with a region-calibration mechanism. Our approach stems from the observation and intuition that standard data augmentations, such as RandomResizedCrop, tend to transform inputs into diverse conditions: easy positives, hard positives, or hard negatives. In traditional distillation frameworks, these transformed samples are utilized equally through their predictive probabilities derived from pretrained teacher models. However, merely relying on prediction values from a pretrained teacher, a common practice in prior studies, neglects the reliability of these soft label predictions. To address this, we propose a new scheme that calibrates the less-confident regions to be the context using softened hard groundtruth labels. Our approach involves the processes of hard regions mining + calibration. We demonstrate empirically that this method can dramatically improve the convergence speed and final accuracy. Additionally, we find that a consistent mixing strategy can stabilize the distributions of soft supervision, taking advantage of the soft labels. As a result, we introduce a stabilized SelfMix augmentation that weakens the variation of the mixed images and corresponding soft labels through mixing similar regions within the same image. FerKD is an intuitive and well-designed learning system that eliminates several heuristics and hyperparameters in former FKD solution. More importantly, it achieves remarkable improvement on ImageNet-1K and downstream tasks. For instance, FerKD achieves 81.2% on ImageNet-1K with ResNet-50, outperforming FKD and FunMatch by remarkable margins. Leveraging better pre-trained weights and larger architectures, our finetuned ViT-G14 even achieves 89.9%. Our code is available at https://github.com/szq0214/FKD/tree/main/FerKD.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes