Abstract
We report a new approach to ONNs based on integrated Kerr micro-combs that is programmable, highly scalable and capable of reaching ultra-high speeds. We demonstrate a single neuron perceptron at 11.9 Giga-OPS at 8 bits per OP, or 95.2 Gbps. We then demonstrate a convolutional accelerator operating beyond 11 TeraOPs/s. We test the perceptron on handwritten-digit recognition and cancer-cell detection-achieving over 90% and 85% accuracy, respectively.
Original language | English |
---|---|
Title of host publication | 2021 Conference on Lasers and Electro-Optics, CLEO 2021 - Proceedings |
Publisher | IEEE, Institute of Electrical and Electronics Engineers |
Number of pages | 2 |
ISBN (Electronic) | 9781943580910 |
DOIs | |
Publication status | Published - 2021 |
Event | Conference on Lasers and Electro-Optics 2021 - Online, San Jose, United States of America Duration: 9 May 2021 → 14 May 2021 https://opg.optica.org/conference.cfm?meetingid=124&yr=2021#AW2E (CLEO: Science and Innovations) https://opg.optica.org/conference.cfm?meetingid=123&yr=2021 (CLEO: Applications and Technology) https://opg.optica.org/conference.cfm?meetingid=155&yr=2021 (CLEO: QELS_Fundamental Science) |
Conference
Conference | Conference on Lasers and Electro-Optics 2021 |
---|---|
Abbreviated title | CLEO 2021 |
Country/Territory | United States of America |
City | San Jose |
Period | 9/05/21 → 14/05/21 |
Internet address |
|