Fermilab Computing Division

> An Overview of Neuromorphic Computing: Models, Architectures and Applications

Full Title: An Overview of Neuromorphic Computing: Models, Architectures and Applications
Date & Time: 01 Sep 2017 at 13:30
Event Location: WH One West
Event Topic(s): Computing Techniques Seminar
Event Moderator(s):
Event Info: Speaker:
Catherine Schuman, Oak Ridge / U. Tennessee

Abstract:
Neuromorphic computers are computers in which their underlying architecture and function is inspired by biological neural systems. Neuromorphic computers promise the ability to do intelligent computing through neural network-style operation, but they do so faster, on lower power and with smaller footprint than traditional computer architectures, making them ideal for deployment into real-world environments and for applications requiring real-time performance. In this talk, an overview of the current state of neuromorphic computing will be presented, including a brief background on neuromorphic models, algorithms, hardware, and applications in the literature. Then, two neuromorphic models will be presented in detail: a fully digital architecture with both custom chip and FPGA-based implementations and a mixed analog-digital architecture with memristor-based synapses. Several applications of these two architectures will be presented to demonstrate the capabilities of neuromorphic systems in solving complex problems. These applications include a high-energy physics data classification application, a real-world anomaly detection application, and an autonomous robot control application.

No talks in agenda


DocDB Home ]  [ Search ] [ Authors ] [ Events ] [ Topics ]

DocDB Version 8.7.23, contact Document Database Administrators
Execution time: 1 wallclock secs ( 0.19 usr + 0.03 sys = 0.22 CPU)