A Study of Parallel Perturbative Gradient Descent
Abstract
We have continued our study of a parallel perturbative learning method [Alspector et al., 1993] and implications for its implemen(cid:173) tation in analog VLSI. Our new results indicate that, in most cases, a single parallel perturbation (per pattern presentation) of the func(cid:173) tion parameters (weights in a neural network) is theoretically the best course. This is not true, however, for certain problems and may not generally be true when faced with issues of implemen(cid:173) tation such as limited precision. In these cases, multiple parallel perturbations may be best as indicated in our previous results.
Cite
Text
Lippe and Alspector. "A Study of Parallel Perturbative Gradient Descent." Neural Information Processing Systems, 1994.Markdown
[Lippe and Alspector. "A Study of Parallel Perturbative Gradient Descent." Neural Information Processing Systems, 1994.](https://mlanthology.org/neurips/1994/lippe1994neurips-study/)BibTeX
@inproceedings{lippe1994neurips-study,
title = {{A Study of Parallel Perturbative Gradient Descent}},
author = {Lippe, D. and Alspector, Joshua},
booktitle = {Neural Information Processing Systems},
year = {1994},
pages = {803-810},
url = {https://mlanthology.org/neurips/1994/lippe1994neurips-study/}
}