NEUROMOPHIC SYSTEMS BASED ON MEMRISTOR AND FERROELECTRIC DEVICES FOR COMPUTING APPLICATIONS

Artificial neural networks (ANNs) are simplified models and hardware that simulate how the brain of living beings, especially humans, process information.

ANNs are inspired by the nervous system and biological behavior, creating a layered interconnection system of artificial neurons that collaborate to process input data and generate output. ANNs differ from other AI models in having the ability to learn automatically (machine learning).

This technology has enormous current impact and potential for future applications, just to mention a few ones: autonomous vehicles, renewable energy, economy and financial predictions, computer visions, and disruptive medicine and health care services.

In the last decade, ANNs have become a very popular topic, but the technology itself has been around for many decades with several problems yet to be solved. For example, current computer systems are highly inadequate in terms of data processing, storage and transmission.

Aspects related to energy consumption are also an unsolved problem, where the rational use of energy and natural resources is a fundamental objective of global sustainability.

NEMERFEC aims to realize novel ANNs through the development of hardware based on memristive and ferroelectric material systems.

Such thin films of binary and complex oxides are to be deposited by physical ablation methods, such as PLD and sputtering, with a high degree of control during deposition.

Advanced characterization techniques, microfabrication and nanofabrication of devices will be performed at the clean room facilities of the Aragon Nanoscience and Materials Institute (INMA-CSIC) to obtain ANNs at nanometric scales and with an ultra-fast response time.

The final ANNs will be implemented in the analysis of thermal images in hyperthermia processes with magnetic nanoparticles at the Nano Scale Biomagnetics company during the non-academic placement period. These possible results will be completely novel in this type of application.