On the Universality of the Double Descent Peak in Ridgeless Regression

Abstract

We prove a non-asymptotic distribution-independent lower bound for the expected mean squared generalization error caused by label noise in ridgeless linear regression. Our lower bound generalizes a similar known result to the overparameterized (interpolating) regime. In contrast to most previous works, our analysis applies to a broad class of input distributions with almost surely full-rank feature matrices, which allows us to cover various types of deterministic or random feature maps. Our lower bound is asymptotically sharp and implies that in the presence of label noise, ridgeless linear regression does not perform well around the interpolation threshold for any of these feature maps. We analyze the imposed assumptions in detail and provide a theory for analytic (random) feature maps. Using this theory, we can show that our assumptions are satisfied for input distributions with a (Lebesgue) density and feature maps given by random deep neural networks with analytic activation functions like sigmoid, tanh, softplus or GELU. As further examples, we show that feature maps from random Fourier features and polynomial kernels also satisfy our assumptions. We complement our theory with further experimental and analytic results.

Published at ICLR 2021. 9 pages + 34 pages appendix. Changes in v8: Small corrections. Experimental results can be reproduced using the code at https://github.com/dholzmueller/universal_double_descent

References in corpus (8)

• On the Convergence of Adam and Beyond
• Understanding deep learning requires rethinking generalization
• Searching for Activation Functions
• Double Trouble in Double Descent : Bias and Variance(s) in the Lazy Regime
• More Data Can Hurt for Linear Regression: Sample-wise Double Descent
• The Neural Tangent Kernel in High Dimensions: Triple Descent and a Multi-Scale Theory of Generalization
• Multiple Descent: Design Your Own Generalization Curve
• Complex powers of analytic functions and meromorphic renormalization in QFT

Cited by in corpus (1)

• Data splitting improves statistical performance in overparametrized regimes