The Research on Neuromorphic Computing & Deep Learning


 The term "neuromorphic computing" was first coined by Carver Mead in the 1980s to mean using electronic systems to mimic the operations in the nervous system. In a more recent and practical sense, the term represents a computing paradigm that mimics the operations in the nervous system to pursue efficient (in terms of area and power consumption) execution of various tasks (such as image/voice recognition) that require some form of intelligence.

"Deep learning" is a branch of machine learning, which attempts to learn how to obtain abstract representations of data by using multiple layers of processing or non-linear transformations.

These two concepts are becoming very important since it seems to be very difficult (if not impossible) to reach the efficiency of a human brain with the traditional computing model based on von Neumann architecture.


This research consists of three sub-topics:

  1. Devices and Circuits for Neuromorphic Computing

  2. Architectures for Neuromorphic Computing and Deep Learning

  3. Deep Learning and Applications

XE Login