Salvato in:
| Autori principali: | , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2509.06696 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915491121987584 |
|---|---|
| author | Sestoft, Joachim E. Jensen, Thomas K. Flodgren, Vidar Das, Abhijit Schlosser, Rasmus D. Alcer, David Lamers, Mariia Kanne, Thomas Borgström, Magnus T. Nygård, Jesper Mikkelsen, Anders |
| author_facet | Sestoft, Joachim E. Jensen, Thomas K. Flodgren, Vidar Das, Abhijit Schlosser, Rasmus D. Alcer, David Lamers, Mariia Kanne, Thomas Borgström, Magnus T. Nygård, Jesper Mikkelsen, Anders |
| contents | Computational hardware designed to mimic biological neural networks holds the promise to resolve the drastically growing global energy demand of artificial intelligence. A wide variety of hardware concepts have been proposed, and among these, photonic approaches offer immense strengths in terms of power efficiency, speed and synaptic connectivity. However, existing solutions have large circuit footprints limiting scaling potential and they miss key biological functions, like inhibition. We demonstrate an artificial nano-optoelectronic neuron with a circuit footprint size reduced by at least a factor of 100 compared to existing technologies and operating powers in the picowatt regime. The neuron can deterministically receive both exciting and inhibiting signals that can be summed and treated with a non-linear function. It demonstrates several biological relevant responses and memory timescales, as well as weighting of input channels. The neuron is compatible with commercial silicon technology, operates at multiple wavelengths and can be used for both computing and optical sensing. This work paves the way for two important research paths: photonic neuromorphic computing with nanosized footprints and low power consumption, and adaptive optical sensing, using the same architecture as a compact, modular front end |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_06696 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Nanoscale photonic neuron with biological signal processing Sestoft, Joachim E. Jensen, Thomas K. Flodgren, Vidar Das, Abhijit Schlosser, Rasmus D. Alcer, David Lamers, Mariia Kanne, Thomas Borgström, Magnus T. Nygård, Jesper Mikkelsen, Anders Mesoscale and Nanoscale Physics Disordered Systems and Neural Networks Computational hardware designed to mimic biological neural networks holds the promise to resolve the drastically growing global energy demand of artificial intelligence. A wide variety of hardware concepts have been proposed, and among these, photonic approaches offer immense strengths in terms of power efficiency, speed and synaptic connectivity. However, existing solutions have large circuit footprints limiting scaling potential and they miss key biological functions, like inhibition. We demonstrate an artificial nano-optoelectronic neuron with a circuit footprint size reduced by at least a factor of 100 compared to existing technologies and operating powers in the picowatt regime. The neuron can deterministically receive both exciting and inhibiting signals that can be summed and treated with a non-linear function. It demonstrates several biological relevant responses and memory timescales, as well as weighting of input channels. The neuron is compatible with commercial silicon technology, operates at multiple wavelengths and can be used for both computing and optical sensing. This work paves the way for two important research paths: photonic neuromorphic computing with nanosized footprints and low power consumption, and adaptive optical sensing, using the same architecture as a compact, modular front end |
| title | Nanoscale photonic neuron with biological signal processing |
| topic | Mesoscale and Nanoscale Physics Disordered Systems and Neural Networks |
| url | https://arxiv.org/abs/2509.06696 |