English

Every spring semester

The purpose of this course is to give an in depth understanding for the physics of common memory device technologies with focus on non-volatile memories. Furthermore, the course covers how these memory devices can be integrated to create neuromorphic hardware for applications in machine learning and artificial intelligence. Finally, the course gives an introduction to the architectures and algorithms that are used in machine learning, to give a basic understanding for the needs that memory devices and their connections need to fulfil.

Memory devices of the computer: SRAM, DRAM, NAND

Non-volatile memory devices: The memristor. Resistive memories (RRAM), phase change memories (PCM), ferroelectric memories (FeRAM), magnetic memories (MRAM).

Integration of memory devices: 3D stacking for scalability, crossbar architecture

Neural network architectures: Feed forward neural networks, recursive networks. spiking neural networks.

Machine learning algorithms: Backpropagation, gradient descent, Hebbian and non-Hebbian learning, unsupervised learning through STDP.

Explain in general the memory hierarchy of a modern computer and in detail how its memory components function.
Explain the physical processes that determine the functionality of common non-volatile memory types such as RRAM, PCM, FeRAM and STT-MRAM.
Understand in general the limitations and benefits of the various memory technologies treated in the course.
Understand in general how the integration of memory devices into neural network circuits can be realized and the benefits and limitations of these approaches.
Understand the structure and function of the basic types of artificial neural networks.
Explain in detail how to perform training of a neural network with the help of backpropagation and the gradient descent algorithm.

perform measurement and analysis of the current-voltage characteristics of a RRAM device.
from a measured polarization-field diagram be able to extract important parameters for a ferroelectric capacitor.
be able to give suggestions on how speed, reliability and energy consumption can be improved in the memory device technologies treated in the course.
design and train a neural network to perform image recognition.

ealize the need for energy efficient and scalable neuromorphic hardware for machine learning and AI.
Evaluate the applicability of a given memory technology for a range of application areas with respect to the pros and cons of the technology.

Lectures

Seminars

Laboratory exercises

Project

Written exam

Written report

Failed, pass

Basic knowledge of device physics

Chen, A., Hutchby, J., Zhirnov, V. & Bourianoff, G.: Emerging Nanoelectronic Devices. John Wiley & Sons, 2015. ISBN 9781118447741.

Course Coordinator: Mattias Borg, mattias.borg@eit.lth.se

EITP25F

2020-10-01

Professor Thomas Johansson