Dynamically Stable Infinite-Width Limits of Neural Classifiers
This work addresses a foundational theoretical problem in machine learning by clarifying the scaling behavior of neural networks, which is incremental but important for researchers in deep learning theory.
The authors tackled the problem of understanding how neural network training behaves in the infinite-width limit by proposing a general framework that unifies and extends existing mean-field and NTK approaches, revealing a finite set of distinct limit models and introducing a novel initialization-corrected mean-field limit that better matches finite-width properties.
Recent research has been focused on two different approaches to studying neural networks training in the limit of infinite width (1) a mean-field (MF) and (2) a constant neural tangent kernel (NTK) approximations. These two approaches have different scaling of hyperparameters with the width of a network layer and as a result, different infinite-width limit models. We propose a general framework to study how the limit behavior of neural models depends on the scaling of hyperparameters with network width. Our framework allows us to derive scaling for existing MF and NTK limits, as well as an uncountable number of other scalings that lead to a dynamically stable limit behavior of corresponding models. However, only a finite number of distinct limit models are induced by these scalings. Each distinct limit model corresponds to a unique combination of such properties as boundedness of logits and tangent kernels at initialization or stationarity of tangent kernels. Existing MF and NTK limit models, as well as one novel limit model, satisfy most of the properties demonstrated by finite-width models. We also propose a novel initialization-corrected mean-field limit that satisfies all properties noted above, and its corresponding model is a simple modification for a finite-width model.