Please use this identifier to cite or link to this item:
http://hdl.handle.net/10071/29488
Author(s): | Miranda, I. D. S. Arora, A. Susskind, Z. Souza, J. S. A. Jadhao, M. P. Villon, L. A. Q. Dutra, D. L. C. Lima, P. M. V. França, F. M. G. Breternitz Jr., M. John, L. K. |
Editor: | Cardoso, J. M. P., Jimborean, A., and Mentens, N. |
Date: | 2023 |
Title: | COIN: Combinational Intelligent Networks |
Book title/volume: | 2023 IEEE 34th International Conference on Application-specific Systems, Architectures and Processors (ASAP) |
Event title: | 34th International Conference on Application-specific Systems, Architectures and Processors (ASAP) |
Reference: | Miranda, I. D. S., Arora, A., Susskind, Z., Souza, J. S. A., Jadhao, M. P., Villon, L. A. Q., Dutra, D. L. C., Lima, P. M. V., França, F. M. G., Breternitz Jr., M., & John, L. K. (2023). COIN: Combinational Intelligent Networks. In J. M. P. Cardoso, A. Jimborean, & N. Mentens (Eds.), 2023 IEEE 34th International Conference on Application-specific Systems, Architectures and Processors (ASAP). IEEE. https://doi.org/10.1109/ASAP57973.2023.00016 |
ISSN: | 2160-0511 |
ISBN: | 979-8-3503-4685-5 |
DOI (Digital Object Identifier): | 10.1109/ASAP57973.2023.00016 |
Keywords: | Weightless neural networks LogicWiSARD Binary neural networks FPGA ASIC |
Abstract: | We introduce Combinational Intelligent Networks (COIN), a machine learning technique that targets edge inference using low-resourced FPGAs or ASICs. COIN is an improvement on LogicWiSARD, a recent weightless neural network that achieves low power, small area, and high throughput. We convert the LogicWiSARD model into a binary neural network, train it using backpropagation, and then convert it to a COIN model. As a result, COIN can achieve higher accuracy than LogicWiSARD or it can require significantly fewer hardware resources when comparing models with similar accuracies. In comparison to a BNN implementation, FINN, small and large COIN models are more energy efficient demonstrating up to 11.5x higher inferences/Joule at similar accuracy. Our tool executes the complete flow, from training to RTL. and is publicly available. |
Peerreviewed: | yes |
Access type: | Embargoed Access |
Appears in Collections: | ISTAR-CRI - Comunicações a conferências internacionais |
Files in This Item:
File | Size | Format | |
---|---|---|---|
conferenceobject_97244.pdf Restricted Access | 198,71 kB | Adobe PDF | View/Open Request a copy |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.