■ Title: Non-volatile Memory for Neuromorphic Computing
■ Speaker: Prof. SangBum Kim (Dept. of MSE, SNU)
■ Date and time: 6/4(Tue) 16:00
■ Venue: Applied Engineering Building (W1) 1st Floor, Multimedia Lecture Room
■ Host : Prof. Byungha Shin
■ Abstract :
Recently, interest in a neuromorphic computing has been growing as one of the new computer architectures suitable for artificial intelligence computation amid competition from global corporations and nations to take the lead in artificial intelligence, which has begun to develop at a rapid pace. In particular, novel computation using emerging memory devices enables a massively parallel computation method, enabling vector and matrix calculations frequently used in machine learning such as deep learning to be executed at high speed and low power. It also demonstrates the possibility of implementing low-power machine learning algorithms through the implementation of spiking neural networks, also called next-generation neural networks. We will introduce the research fields which are organized in materials, devices, circuits, architectures and algorithms in order to realize neuromorphic computing using emerging memory that can express synaptic weights.