Memristive Devices for Neuromorphic Applications: Comparative Analysis
- PDF / 614,240 Bytes
- 14 Pages / 595.276 x 790.866 pts Page_size
- 66 Downloads / 235 Views
Memristive Devices for Neuromorphic Applications: Comparative Analysis Victor Erokhin 1,2 Accepted: 2 October 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Neuromorphic systems must have at least five unavoidable features that are present in living beings. First, neuromorphic systems must perform memorizing and processing functions, using same elements. Second, it must allow data acquisition from sensors with preliminary processing and recording. Third, it must allow non-equilibrium processes, such as oscillator behavior at fixed values of input stimuli. Fourth, it must mimic some features of nervous systems. Fifth, it must permit possibility of coupling with living beings. In this paper, these features are considered with a special attention to how memristive devices can be implemented for reaching the goal. Comparison of characteristic properties of organic and inorganic memristive devices is discussed. Keywords Neuromorphic systems . Memristive devices . Memory and processing . Oscillators . Sensors
1 Introduction An explosive activity in the field of memristive device [1] applicability for neuromorphic systems during last decade is due to their capabilities of mimicking some properties of elements of nervous systems in artificial circuits and networks. It is possible to distinguish such applications as memory, artificial neural networks (ANN), sensors, oscillators, nervous system mimicking circuits, interfacing with living beings, which can result in the realization of prosthesis of some parts of the nervous system. These applications are strongly connected between each other. Same elements can be involved in performing different functions (for example: memorizing and processing), and therefore, it is possible to make only approximate division of tasks. Usually, neuromorphic applications of memristive devices are connected to the possibilities of mimicking neuron and synapse properties in artificial circuits and systems, as it is indicated in reviews, dedicated to inorganic [2–6] and organic [7] memristive devices. In this short review, I will make some remarks regarding the areas of the applications of memristive devices for
* Victor Erokhin [email protected] 1
IMEM-CNR, Institute of Materials for Electronics and Magnetism, Parco Area delle Scienze, 37A, 43124 Parma, Italy
2
Natinal Research Centre “Kurchatov Institute”, Akademika Kurchatova square 1, Moscow, Russian Federation 123182
neuromorphic systems. Advantages and drawbacks of inorganic and organic systems for each application, mentioned above, will be also considered. The most of inorganic memristive devices are based on sandwich structures, where thin insulating layer is placed between two metal electrodes. In most of the cases, the insulator layer is fabricated from metal oxide materials [8–12], deposited by different techniques. However, some other inorganic materials were also used for the memristive device fabrication, such as, for example, ferroelectrics [13, 14] and silicon nanostructures [15]
Data Loading...
