Model Reprogramming Demystified: A Neural Tangent Kernel Perspective
This provides theoretical foundations for MR, addressing a gap for researchers and practitioners using this efficient adaptation technique, though it is incremental as it builds on existing NTK theory.
The paper tackles the lack of theoretical understanding of Model Reprogramming (MR), a resource-efficient method for adapting pre-trained models to new tasks, by analyzing it through the Neural Tangent Kernel (NTK) framework. They show that MR success depends on the NTK eigenvalue spectrum on target data and the source model's effectiveness, with experiments validating these insights.
Model Reprogramming (MR) is a resource-efficient framework that adapts large pre-trained models to new tasks with minimal additional parameters and data, offering a promising solution to the challenges of training large models for diverse tasks. Despite its empirical success across various domains such as computer vision and time-series forecasting, the theoretical foundations of MR remain underexplored. In this paper, we present a comprehensive theoretical analysis of MR through the lens of the Neural Tangent Kernel (NTK) framework. We demonstrate that the success of MR is governed by the eigenvalue spectrum of the NTK matrix on the target dataset and establish the critical role of the source model's effectiveness in determining reprogramming outcomes. Our contributions include a novel theoretical framework for MR, insights into the relationship between source and target models, and extensive experiments validating our findings.