On the Mechanisms of Weak-to-Strong Generalization: A Theoretical Perspective
This addresses a fundamental gap in machine learning theory for researchers, but it is incremental as it builds on widely observed phenomena with theoretical insights.
The paper tackles the problem of understanding weak-to-strong generalization, where a student model outperforms a weaker teacher, by theoretically analyzing simple models to uncover three core mechanisms, such as compensation for under-regularization and leveraging broader pre-training, resulting in proofs of lower test error and outperformance.
Weak-to-strong generalization, where a student model trained on imperfect labels generated by a weaker teacher nonetheless surpasses that teacher, has been widely observed but the mechanisms that enable it have remained poorly understood. In this paper, through a theoretical analysis of simple models, we uncover three core mechanisms that can drive this phenomenon. First, by analyzing ridge regression, we study the interplay between the teacher and student regularization and prove that a student can compensate for a teacher's under-regularization and achieve lower test error. We also analyze the role of the parameterization regime of the models. Second, by analyzing weighted ridge regression, we show that a student model with a regularization structure more aligned to the target, can outperform its teacher. Third, in a nonlinear multi-index setting, we demonstrate that a student can learn easy, task-specific features from the teacher while leveraging its own broader pre-training to learn hard-to-learn features that the teacher cannot capture.