Almost Linear VC-Dimension Bounds for Piecewise Polynomial Networks-Reference-Cited by-同舟云学术

Almost Linear VC-Dimension Bounds for Piecewise Polynomial Networks

Published:1998-11-01 Issue:8 Volume:10 Page:2159-2173
ISSN:0899-7667
Container-title:Neural Computation
language:en
Short-container-title:Neural Computation

Author:

Bartlett Peter L.¹,Maiorov Vitaly²,Meir Ron³

Affiliation:

1. Department of System Engineering, Australian National University, Canberra, ACT 0200, Australia

2. Department of Mathematics, Technion, Haifa 32000, Israel

3. Department of Electrical Engineering, Technion, Haifa 32000, Israel

Abstract

We compute upper and lower bounds on the VC dimension and pseudodimension of feedforward neural networks composed of piecewise polynomial activation functions. We show that if the number of layers is fixed, then the VC dimension and pseudo-dimension grow as W log W, where W is the number of parameters in the network. This result stands in opposition to the case where the number of layers is unbounded, in which case the VC dimension and pseudo-dimension grow as W2. We combine our results with recently established approximation error rates and determine error bounds for the problem of regression estimation by piecewise polynomial networks with unbounded weights.

Publisher

MIT Press - Journals

Subject

Cognitive Neuroscience,Arts and Humanities (miscellaneous)

Link

https://www.mitpressjournals.org/doi/pdf/10.1162/089976698300017016

Reference15 articles.

1. Approximation and estimation bounds for artificial neural networks

2. The sample complexity of pattern classification with neural networks: the size of the weights is more important than the size of the network

3. Fat-Shattering and the Learnability of Real-Valued Functions

4. Nonparametric Maximum Likelihood Estimation by the Method of Sieves

5. Bounding the Vapnik-Chervonenkis dimension of concept classes parameterized by real numbers

Cited by 49 articles. 订阅此论文施引文献订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献

1. Estimating a regression function in exponential families by model selection;Bernoulli;2024-05-01

2. To understand double descent, we need to understand VC theory;Neural Networks;2024-01

3. A Deep Neural Network Regularization Measure: The Class-Based Decorrelation Method;Entropy;2023-12-20

4. Approximation of classifiers by deep perceptron networks;Neural Networks;2023-08

5. Spectrahedral Regression;SIAM Journal on Optimization;2023-06-01