Theoretical evaluation of the computational capabilities of a random network with memristive connections
Creating computing systems that implement the information processing properties of biological brains can have several advantages in terms of computational and energy efficiency. The present work provides a numerical modeling tool to evaluate the computational capabilities of a neuromorphic hardware...
- Autores:
-
Suárez Usme, Laura Estefany
- Tipo de recurso:
- Fecha de publicación:
- 2016
- Institución:
- Universidad de los Andes
- Repositorio:
- Séneca: repositorio Uniandes
- Idioma:
- eng
- OAI Identifier:
- oai:repositorio.uniandes.edu.co:1992/13855
- Acceso en línea:
- http://hdl.handle.net/1992/13855
- Palabra clave:
- Redes neurales (Neurobiología) - Investigaciones
Sistemas dinámicos - Investigaciones
Análisis numérico - Investigaciones
Ingeniería
- Rights
- openAccess
- License
- https://repositorio.uniandes.edu.co/static/pdf/aceptacion_uso_es.pdf
| Summary: | Creating computing systems that implement the information processing properties of biological brains can have several advantages in terms of computational and energy efficiency. The present work provides a numerical modeling tool to evaluate the computational capabilities of a neuromorphic hardware architecture consisting of a random network of memristive connections. This architecture is analyzed from the perspective of the Reservoir Computing approach. A simple methodology based on the definition of ordered and chaotic dynamical systems was used to determine the Separation and Fading Memory Properties of the architecture proposed as required by the aforementioned approach. Results show the potential use of these networks as reservoirs for the processing of time-varying inputs. |
|---|