Robust Fine-tuning for Pre-trained 3D Point Cloud Models

TL;DR

This paper introduces WiSE-FT-LP, a robust fine-tuning method for pre-trained 3D point cloud models that improves feature robustness under distribution shifts without changing model structures.

Contribution

The paper proposes WiSE-FT-LP, a novel weight-space ensemble approach that enhances robustness of 3D point cloud models during fine-tuning and linear probing.

Findings

WiSE-FT-LP improves robustness under distribution shifts.

The method maintains high downstream task performance.

It effectively balances robustness and accuracy.

Abstract

This paper presents a robust fine-tuning method designed for pre-trained 3D point cloud models, to enhance feature robustness in downstream fine-tuned models. We highlight the limitations of current fine-tuning methods and the challenges of learning robust models. The proposed method, named Weight-Space Ensembles for Fine-Tuning then Linear Probing (WiSE-FT-LP), integrates the original pre-training and fine-tuning models through weight space integration followed by Linear Probing. This approach significantly enhances the performance of downstream fine-tuned models under distribution shifts, improving feature robustness while maintaining high performance on the target distribution. We apply this robust fine-tuning method to mainstream 3D point cloud pre-trained models and evaluate the quality of model parameters and the degradation of downstream task performance. Experimental results demonstrate the effectiveness of WiSE-FT-LP in enhancing model robustness, effectively balancing downstream task performance and model feature robustness without altering the model structures.