Placing trainable nonlinear functions on connections in analogue networks enables efficient representation of smooth continuous targets with hardware transfer at projected 30 microwatt power.
G., Onodera, T., Stein, M
6 Pith papers cite this work, alongside 704 external citations. Polarity classification is still indexing.
6
Pith papers citing it
704
external citations · OpenAlex
citation-role summary
background 3
citation-polarity summary
years
2026 6roles
background 3polarities
background 3representative citing papers
Spatial multiplexing in optical neural networks is repurposed as a trainable representational coordinate, demonstrated in multi-layer architectures for image classification, regression, and hybrid vision-language captioning with over one million optical phase parameters.
CNN and hand-crafted feature networks recover magnetic field (~3.8 µT), temperature (~0.12 K), and hysteresis branch from one quantitative magneto-optical domain map of Bi:YIG.
Score-based diffusion built intrinsically on the quantum pure-state manifold CP^{d-1}, trained with a local-time Gaussian teacher, matches pure-state ensembles far better than Euclidean baselines in the local-cluster regime, with gains shrinking on globally spread ensembles.
Physical neural substrates realize inference and adaptation via native physics and occupy complementary regimes; no single platform dominates the proposed static/dynamic benchmarks.
A simulated all-optical CNN achieves ~94% MNIST accuracy using a digital twin for pretraining and SPSA for in-situ fine-tuning, but hardware is not realized and key numbers are inconsistent.
citing papers explorer
-
Low-power analogue neural networks with trainable nonlinear connections for continuous control
Placing trainable nonlinear functions on connections in analogue networks enables efficient representation of smooth continuous targets with hardware transfer at projected 30 microwatt power.
-
Multi-channel Optical Vision Model
Spatial multiplexing in optical neural networks is repurposed as a trainable representational coordinate, demonstrated in multi-layer architectures for image classification, regression, and hybrid vision-language captioning with over one million optical phase parameters.
-
Decoding magnetic texture
CNN and hand-crafted feature networks recover magnetic field (~3.8 µT), temperature (~0.12 K), and hysteresis branch from one quantitative magneto-optical domain map of Bi:YIG.
-
Local-Time Riemannian Score Matching on the Quantum Pure-State Manifold
Score-based diffusion built intrinsically on the quantum pure-state manifold CP^{d-1}, trained with a local-time Gaussian teacher, matches pure-state ensembles far better than Euclidean baselines in the local-cluster regime, with gains shrinking on globally spread ensembles.
-
Beyond Silicon: Materials, Mechanisms, and Methods for Physical Neural Computing
Physical neural substrates realize inference and adaptation via native physics and occupy complementary regimes; no single platform dominates the proposed static/dynamic benchmarks.
-
Photonic convolutional neural network with pre-trained in situ training
A simulated all-optical CNN achieves ~94% MNIST accuracy using a digital twin for pretraining and SPSA for in-situ fine-tuning, but hardware is not realized and key numbers are inconsistent.