研究设定与方法
We study two-layer neural networks trained by stochastic gradient descent (SGD) on a multi-index teacher model, focusing on how first-layer parameters orthogonal to the teacher subspace affect generalization. We decompose the first-layer weights into a signal component aligned with the teacher and an orthogonal complement, and analyze their joint SGD dynamics together with the second layer in the proportional limit.
核心动态发现
We show that although SGD updates to the complement are initially isotropic, the training dynamics induce teacher-dependent spectral spikes in the complement Gram matrix. This emergent anisotropy provides a concrete mechanism by which representation variance is amplified, and generalization degrades.
分析工具与实验验证
To isolate this effect, we introduce Decomposed Dynamics, an analytically tractable surrogate whose asymptotic test-error dynamics coincide with those of SGD. In simulations with a ReLU student, freezing the directions of the orthogonal complement improves generalization relative to training them.
补充结果与结论
Finally, we show that the signal component rapidly aligns with the second-layer weights, revealing strong cross-layer coupling even in this shallow setting. Together, our results provide a dynamical explanation for how feature learning outside the teacher subspace harms generalization.