Probabilistic photonic computing with chaotic light
Biological neural networks effortlessly tackle complex computational problems and excel at predicting outcomes from noisy, incomplete data. Artificial neural networks (ANNs), inspired by these biological counterparts, have emerged as powerful tools for deciphering intricate data patterns and making predictions. However, conventional ANNs can be viewed as "point estimates" that do not capture the uncertainty of prediction, which is an inherently probabilistic process. In contrast, treating an ANN as a probabilistic model derived via Bayesian inference poses significant challenges for conventional deterministic computing architectures. Here, we use chaotic light in combination with incoherent photonic data processing to enable high-speed probabilistic computation and uncertainty quantification. We exploit the photonic probabilistic architecture to simultaneously perform image classification and uncertainty prediction via a Bayesian neural network. Our prototype demonstrates the seamless cointegration of a physical entropy source and a computational architecture that enables ultrafast probabilistic computation by parallel sampling.
Authors
- C. David Wright (ORCID: https://orcid.org/0000-0003-4087-7467)
- Benjamin Risse (ORCID: https://orcid.org/0000-0001-5691-4029)
- Bernhard Klein (ORCID: https://orcid.org/0000-0003-0497-5748)
- Holger Fröning (ORCID: https://orcid.org/0000-0001-9562-0680)
- Wolfram H. P. Pernice (ORCID: https://orcid.org/0000-0003-4569-4213)
- Harish Bhaskaran (ORCID: https://orcid.org/0000-0003-0774-8110)
- Frank Brückerhoff‐Plückelmann (ORCID: https://orcid.org/0009-0006-5748-7069)
- Martin Salinga (ORCID: https://orcid.org/0000-0002-2228-6244)
- Jelle Dijkstra
- Marlon Becker
- Hendrik Borras (ORCID: https://orcid.org/0000-0002-2411-2416)
- Martin Brückerhoff
- Akhil Varri (ORCID: https://orcid.org/0009-0008-5948-4181)
Institutions
- Heidelberg University (DE)
- University of Münster (DE)
- University of Exeter (GB)
- University of Oxford (GB)
- Kirchhoff (Germany) (DE)
- Stadtwerke Köln (Germany) (DE)
- FH Münster University of Applied Sciences (DE)
Publication Details
- Journal
- Nature Communications
- Published
- 2024-12-01
- DOI
- https://doi.org/10.1038/s41467-024-54931-6
- Citations
- 23
- Primary Topic
- Neural Networks and Reservoir Computing
- Type
- article
- Field-Weighted Citation Impact
- 7.99
Funders
- European Commission
- Österreichische Forschungsförderungsgesellschaft
- Engineering and Physical Sciences Research Council