Reconfigurable Neural Net Chip with 32k Connections
Abstract
We describe a CMOS neural net chip with a reconfigurable network archi(cid:173) tecture. It contains 32,768 binary, programmable connections arranged in 256 'building block' neurons. Several 'building blocks' can be connected to form long neurons with up to 1024 binary connections or to form neurons with analog connections. Single- or multi-layer networks can be imple(cid:173) mented with this chip. We have integrated this chip into a board system together with a digital signal processor and fast memory. This system is currently in use for image processing applications in which the chip extracts features such as edges and corners from binary and gray-level images.
Cite
Text
Graf et al. "Reconfigurable Neural Net Chip with 32k Connections." Neural Information Processing Systems, 1990.Markdown
[Graf et al. "Reconfigurable Neural Net Chip with 32k Connections." Neural Information Processing Systems, 1990.](https://mlanthology.org/neurips/1990/graf1990neurips-reconfigurable/)BibTeX
@inproceedings{graf1990neurips-reconfigurable,
title = {{Reconfigurable Neural Net Chip with 32k Connections}},
author = {Graf, H. P. and Janow, R. and Henderson, D. and Lee, R.},
booktitle = {Neural Information Processing Systems},
year = {1990},
pages = {1032-1038},
url = {https://mlanthology.org/neurips/1990/graf1990neurips-reconfigurable/}
}