Acquisition and Representation of Spatio-Temporal Signals in Polychronizing Spiking Neural Networks
- Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
The ability of an intelligent agent to process complex signals such as those found in audio or video depends heavily on the nature of the internal representation of the relevant information. Our report explores the mechanisms underlying this process by investigating theories inspired by the function of the neocortex. Specifically, we focus on the phenomenon of polychronization, which describes the self-organization in a spiking neural network resulting from the interplay between network structure, driven spiking activity, and synaptic plasticity. What emerges are groups of neurons that exhibit reproducible, time-locked patterns of spiking activity. We propose that this representation is well suited to spatio-temporal signal processing, as it naturally resembles patterns found in real-world signals. We explore the computational properties of this method and demonstrate the ability of a simple polychronizing network to learn different spatio-temporal signals.
- Research Organization:
- Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
- Sponsoring Organization:
- USDOE National Nuclear Security Administration (NNSA)
- DOE Contract Number:
- AC04-94AL85000; NA0003525
- OSTI ID:
- 1575266
- Report Number(s):
- SAND-2019-4971J; 675222
- Resource Relation:
- Conference: 7. Proceedings of the Annual Neuro-inspired Computational Elements Workshop (NICE '19), Albany, NY (United States), 26-28 Mar 2019
- Country of Publication:
- United States
- Language:
- English
Similar Records
Skip-Connected Self-Recurrent Spiking Neural Networks with Joint Intrinsic Parameter and Synaptic Weight Training
Temporally coordinated spiking activity of human induced pluripotent stem cell-derived neurons co-cultured with astrocytes