On consequences of finetuning on data with highly discriminative features

Wojciech Masarczyk, Tomasz Trzcinski, Mateusz Ostaszewski

Keywords: finetuning, hidden represntations, transfer learning

Abstract:

In the era of transfer learning, training neural networks from scratch is becoming obsolete. Transfer learning leverages prior knowledge for new tasks, conserving computational resources. While its advantages are well-documented, we uncover a notable drawback: networks tend to prioritize basic data patterns, forsaking valuable pre-learned features. We term this behavior "feature erosion" and analyze its impact on network performance and internal representations.

Track: Extended Abstract Track

Submission Number: 54