TY - JOUR
T1 - Random vector functional link networks for function approximation on manifolds
AU - Needell, D
AU - Nelson, AA
AU - Saab, R
AU - Salanevich, P
AU - Schavemaker, O
N1 - Publisher Copyright:
Copyright © 2024 Needell, Nelson, Saab, Salanevich and Schavemaker.
PY - 2024/4/17
Y1 - 2024/4/17
N2 - The learning speed of feed-forward neural networks is notoriously slow and has presented a bottleneck in deep learning applications for several decades. For instance, gradient-based learning algorithms, which are used extensively to train neural networks, tend to work slowly when all of the network parameters must be iteratively tuned. To counter this, both researchers and practitioners have tried introducing randomness to reduce the learning requirement. Based on the original construction of Igelnik and Pao, single layer neural-networks with random input-to-hidden layer weights and biases have seen success in practice, but the necessary theoretical justification is lacking. In this study, we begin to fill this theoretical gap. We then extend this result to the non-asymptotic setting using a concentration inequality for Monte-Carlo integral approximations. We provide a (corrected) rigorous proof that the Igelnik and Pao construction is a universal approximator for continuous functions on compact domains, with approximation error squared decaying asymptotically like O(1/n) for the number n of network nodes. We then extend this result to the non-asymptotic setting, proving that one can achieve any desired approximation error with high probability provided n is sufficiently large. We further adapt this randomized neural network architecture to approximate functions on smooth, compact submanifolds of Euclidean space, providing theoretical guarantees in both the asymptotic and non-asymptotic forms. Finally, we illustrate our results on manifolds with numerical experiments.
AB - The learning speed of feed-forward neural networks is notoriously slow and has presented a bottleneck in deep learning applications for several decades. For instance, gradient-based learning algorithms, which are used extensively to train neural networks, tend to work slowly when all of the network parameters must be iteratively tuned. To counter this, both researchers and practitioners have tried introducing randomness to reduce the learning requirement. Based on the original construction of Igelnik and Pao, single layer neural-networks with random input-to-hidden layer weights and biases have seen success in practice, but the necessary theoretical justification is lacking. In this study, we begin to fill this theoretical gap. We then extend this result to the non-asymptotic setting using a concentration inequality for Monte-Carlo integral approximations. We provide a (corrected) rigorous proof that the Igelnik and Pao construction is a universal approximator for continuous functions on compact domains, with approximation error squared decaying asymptotically like O(1/n) for the number n of network nodes. We then extend this result to the non-asymptotic setting, proving that one can achieve any desired approximation error with high probability provided n is sufficiently large. We further adapt this randomized neural network architecture to approximate functions on smooth, compact submanifolds of Euclidean space, providing theoretical guarantees in both the asymptotic and non-asymptotic forms. Finally, we illustrate our results on manifolds with numerical experiments.
KW - feed-forward neural networks
KW - function approximation
KW - machine learning
KW - random vector functional link
KW - smooth manifold
UR - https://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=d7dz6a2i7wiom976oc9ff2iqvdhv8k5x&SrcAuth=WosAPI&KeyUT=WOS:001210857900001&DestLinkType=FullRecord&DestApp=WOS_CPL
UR - http://www.scopus.com/inward/record.url?scp=85191789120&partnerID=8YFLogxK
U2 - 10.3389/fams.2024.1284706
DO - 10.3389/fams.2024.1284706
M3 - Article
SN - 2297-4687
VL - 10
JO - Frontiers in Applied Mathematics and Statistics
JF - Frontiers in Applied Mathematics and Statistics
M1 - 1284706
ER -