Benjamin Scellier

Benjamin Scellier

Auto-generated profile, based on publicly available information

Benjamin Scellier

Rain AI , San Francisco, California, USA

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
3 September 2025 in Nature
Verified
Journal Article
Closed access
PDF file
AMAli Momeni
Ali Momeni
BRBabak Rahmani
Babak Rahmani
BSBenjamin Scellier
Benjamin Scellier
LWLogan G. Wright
Logan G. Wright
PMPeter L. McMahon
Peter L. McMahon
RFRomain Fleury
Romain Fleury

13 April 2023 in Physical Review Research
Verified
Journal Article
Open access
PDF file
VAVidyesh Rao Anisetti
Vidyesh Rao Anisetti
BSB. Scellier
B. Scellier
JSJ. M. Schwarz
J. M. Schwarz
Both non-neural and neural biological systems can learn. So rather than focusing on purely brain-like learning, efforts are underway to study learning in physical systems. Such efforts include equilibrium propagation (EP) and coupled learning (CL), which require storage of two different states—the free state and the perturbed state—during the learning process to retain information about gradients. Here, we propose a learning algorithm rooted in chemical signaling that does not require storage of two different states. Rather, the output error information is encoded in a chemical signal that diffuses into the network in a similar way as the activation/feedforward signal. The steady-state feedback chemical concentration, along with the activation signal, stores the required gradient information locally. We apply our algorithm using a physical, linear flow network and test it using the Iris data set with 93% accuracy. We also prove that our algorithm performs gradient descent. Finally, in addition to comparing our algorithm directly with EP and CL, we address the biological plausibility of the algorithm.

3 April 2025 in Physical Review Applied
Verified
Journal Article
Closed access
BSBenjamin Scellier
Benjamin Scellier
SMSiddhartha Mishra
Siddhartha Mishra

23 June 2022 in arXiv
Verified
Preprint
Open access
PDF file
VAVidyesh Rao Anisetti
Vidyesh Rao Anisetti
BSB. Scellier
B. Scellier
JSJ. M. Schwarz
J. M. Schwarz
Both non-neural and neural biological systems can learn. So rather than focusing on purely brain-like learning, efforts are underway to study learning in physical systems. Such efforts include equilibrium propagation (EP) and coupled learning (CL), which require storage of two different states-the free state and the perturbed state-during the learning process to retain information about gradients. Inspired by slime mold, we propose a new learning algorithm rooted in chemical signaling that does not require storage of two different states. Rather, the output error information is encoded in a chemical signal that diffuses into the network in a similar way as the activation/feedforward signal. The steady state feedback chemical concentration, along with the activation signal, stores the required gradient information locally. We apply our algorithm using a physical, linear flow network and test it using the Iris data set with 93% accuracy. We also prove that our algorithm performs gradient descent. Finally, in addition to comparing our algorithm directly with EP and CL, we address the biological plausibility of the algorithm.

1 June 2026 in APL Quantum
Verified
Journal Article
Open access
BSBenjamin Scellier
Benjamin Scellier
Equilibrium propagation (EP) is a training framework for physical systems that minimize an energy function. EP uses the system’s intrinsic physics during both inference and training, making it a candidate for the development of energy-efficient processors for machine learning. EP has been studied in various classical physical systems, including classical Ising networks and elastic networks. Here, we derive a version of EP for quantum systems in which the Hamiltonian’s expectation value serves as the energy function, with the minimum corresponding to the ground state. As illustrations, we consider the transverse-field Ising network and the quantum harmonic oscillator network, which are quantum analogs of the network models studied within classical EP.