Research Area:  Machine Learning

Abstract:

Transfer learning, also referred as knowledge transfer, aims at reusing knowledge from a source dataset to a similar target one. While several studies address the problem of what to transfer, the very important question of when to answer remains mostly unanswered, especially from a theoretical point-of-view for regression problems. A new theoretical framework for the problem of parameter transfer for the linear model is proposed. It is shown that the quality of transfer for a new input vector depends on its representation in an eigenbasis involving the parameters of the problem. Furthermore, a statistical test is constructed to predict whether a fine-tuned model has a lower prediction quadratic risk than the base target model for an unobserved sample. Efficiency of the test is illustrated on synthetic data as well as real electricity consumption data.

Keywords:  
Transfer learning
Source dataset
Regression problems
Quadratic risk

Author(s) Name:  David Obst, Badih Ghattas, Sandra Claudel

Publisher name:  Elsevier

DOI:  10.1016/j.csda.2022.107499

Volume Information:  Volume 174

Paper Link:   https://www.sciencedirect.com/science/article/abs/pii/S0167947322000792