Surprises in High-Dimensional Ridgeless Least Squares Interpolation
Hastie, Trevor ; Montanari, Andrea ; Rosset, Saharon ; Tibshirani, Ryan J.
arXiv, Tome 2019 (2019) no. 0, / Harvested from
Interpolators---estimators that achieve zero training error---have attracted growing attention in machine learning, mainly because state-of-the art neural networks appear to be models of this type. In this paper, we study minimum $\ell_2$ norm ("ridgeless") interpolation in high-dimensional least squares regression. We consider two different models for the feature distribution: a linear model, where the feature vectors $x_i \in \mathbb{R}^p$ are obtained by applying a linear transform to a vector of i.i.d. entries, $x_i = \Sigma^{1/2} z_i$ (with $z_i \in \mathbb{R}^p$); and a nonlinear model, where the feature vectors are obtained by passing the input through a random one-layer neural network, $x_i = \varphi(W z_i)$ (with $z_i \in \mathbb{R}^d$, $W \in \mathbb{R}^{p \times d}$ a matrix of i.i.d. entries, and $\varphi$ an activation function acting componentwise on $W z_i$). We recover---in a precise quantitative way---several phenomena that have been observed in large-scale neural networks and kernel machines, including the "double descent" behavior of the prediction risk, and the potential benefits of overparametrization.
Publié le : 2019-03-19
Classification:  Mathematics - Statistics Theory,  Computer Science - Machine Learning,  Statistics - Machine Learning
@article{1903.08560,
     author = {Hastie, Trevor and Montanari, Andrea and Rosset, Saharon and Tibshirani, Ryan J.},
     title = {Surprises in High-Dimensional Ridgeless Least Squares Interpolation},
     journal = {arXiv},
     volume = {2019},
     number = {0},
     year = {2019},
     language = {en},
     url = {http://dml.mathdoc.fr/item/1903.08560}
}
Hastie, Trevor; Montanari, Andrea; Rosset, Saharon; Tibshirani, Ryan J. Surprises in High-Dimensional Ridgeless Least Squares Interpolation. arXiv, Tome 2019 (2019) no. 0, . http://gdmltest.u-ga.fr/item/1903.08560/