Information Technology Laboratory, Applied and Computational Mathematics Division
NIST only participates in the February and August reviews.
| name |
email |
phone |
|
| Andrew Martin Dienstfrey |
andrew.dienstfrey@nist.gov |
303.497.7485 |
Physics-based computation is a branch of unconventional computing which uses physical systems evolving by their native dynamics to process information in a controlled way. Our research covers a broad range of areas from theory, to algorithmic, to computational prototypes in novel hardware. At the foundational level we are interested in explorations of the fundamental computational capacity of dynamical systems, for example, spiking-neural networks. Additionally, we are interested in the fundamental limits of computation arising from statistical mechanics and thermodynamic considerations. Research and development of hardware-aware, fault-tolerant training strategies that can be used for in-situ training of neuromorphic hardware are examples of our algorithmic work [1], [2]. At the hardware level, we have ongoing collaborations with NIST and external research groups developing novel devices for next-generation neuromorphic computing including, for example, resistive crossbar structures [3], superconducting optoelectronics [4], and Josephson Junctions [5]. We look forward to having you join us in this exciting area.
[1] A. N. McCaughan, B. G. Oripov, N. Ganesh, S. W. Nam, A. Dienstfrey, S. M. Buckley, "Multiplexed gradient descent: Fast online training of modern datasets on hardware neural networks without backpropagation," APL Machine Learning, v1#2 (2023).
[2] J. Zhao, S. Huang, O. Yousuf, Y. Gao, B. D. Hoskins, G. C. Adam, "Gradient decomposition methods for training neural networks with non-ideal synaptic devices," Frontiers in Neuroscience, v15 (2021)
[3] O. Yousuf, I. Hossen, M. W. Daniels, M. Leuker-Boden, A. Dienstfrey, G. C. Adam, "Device Modeling Bias in ReRAM-Based Neural Network Simulations", IEEE Journal on Emerging and Selected Topics in Circuits and Systems, v13#1 (2023)
[4] S. Khan, B. A. Primavera, J. Chiles, A. N. McCaughan, S. M. Buckley, A. N. Tait, et. al., "Superconducting optoelectronic single-photon synapses," Nature Electronics, v5#10 (2022)
[5] M. L. Schneider, E. M. Jué, M. R. Pufall, K. Segall, C. W. Anderson, "Self-training superconducting neuromorphic circuits using reinforcement learning rules," arXiv:2404.18774 (2024)
Neuromorphic computing; Thermodynamics of computation; Artificial intelligence; Machine learning; Spiking neural networks; Reservoir computing