🤖 AI Summary
This paper addresses robustness in transfer learning under unreliable source data—specifically, the risk of negative transfer arising from classifier ambiguity in Bayesian settings and weak transferable signals between source and target distributions. To tackle this, we propose the *Ambiguity Degree*—a theoretical measure quantifying classification uncertainty—and the *Transfer Around Boundaries* (TAB) framework. TAB employs an adaptive thresholding mechanism to dynamically balance source- and target-domain performance, achieving, for the first time in nonparametric classification and logistic regression, risk-guaranteed transfer learning with provable improvement. Theoretical analysis establishes an approximation-optimal upper bound on excess error—tight up to at most a logarithmic factor. Extensive experiments on synthetic and real-world datasets demonstrate that TAB significantly improves classification accuracy and enhances robustness against distributional shifts and source-data corruption.
📝 Abstract
This paper addresses challenges in robust transfer learning stemming from ambiguity in Bayes classifiers and weak transferable signals between the target and source distribution. We introduce a novel quantity called the ''ambiguity level'' that measures the discrepancy between the target and source regression functions, propose a simple transfer learning procedure, and establish a general theorem that shows how this new quantity is related to the transferability of learning in terms of risk improvements. Our proposed ''Transfer Around Boundary'' (TAB) model, with a threshold balancing the performance of target and source data, is shown to be both efficient and robust, improving classification while avoiding negative transfer. Moreover, we demonstrate the effectiveness of the TAB model on non-parametric classification and logistic regression tasks, achieving upper bounds which are optimal up to logarithmic factors. Simulation studies lend further support to the effectiveness of TAB. We also provide simple approaches to bound the excess misclassification error without the need for specialized knowledge in transfer learning.
