Reservoir computing in materio: A computational framework for in materio computing

Matthew Dale; Susan Stepney; Julian Francis Miller; Martin Albrecht Trefzer

doi:10.1109/IJCNN.2017.7966119

Reservoir computing in materio: A computational framework for in materio computing

Matthew Dale, Susan Stepney, Julian Francis Miller, Martin Albrecht Trefzer

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

The Reservoir Computing (RC) framework is said to have the potential to transfer onto any input-driven dynamical system, provided two properties are present: (i) a fading memory, and (ii) input separability. A typical reservoir consists of a fixed network of recurrently connected processing units; however recent hardware implementations have shown reservoirs are not ultimately bound by this architecture. Previously, we have demonstrated how the RC framework can be applied to randomly-formed carbon nanotube composites to solve computational tasks. Here, we apply the RC framework to an evolvable substrate and compare performance to an already established in materia training technique, referred to as evolution in materia. The results show that by adding the programmable reservoir layer, reservoir computing in materia can significantly outperform the original evolution in materia implementation. This suggests the RC framework offers improved performance, even across non-temporal tasks, when combined with the evolution in materia technique.

Original language	English
Title of host publication	International Joint Conference on Neural Networks (IJCNN)
Publisher	IEEE
Pages	2178-2185
DOIs	https://doi.org/10.1109/IJCNN.2017.7966119
Publication status	Published - 3 Jul 2017

Access to Document

10.1109/IJCNN.2017.7966119

Cite this

@inproceedings{a542479c23d94a8baad98b4c160373ad,

title = "Reservoir computing in materio: A computational framework for in materio computing",

abstract = "The Reservoir Computing (RC) framework is said to have the potential to transfer onto any input-driven dynamical system, provided two properties are present: (i) a fading memory, and (ii) input separability. A typical reservoir consists of a fixed network of recurrently connected processing units; however recent hardware implementations have shown reservoirs are not ultimately bound by this architecture. Previously, we have demonstrated how the RC framework can be applied to randomly-formed carbon nanotube composites to solve computational tasks. Here, we apply the RC framework to an evolvable substrate and compare performance to an already established in materia training technique, referred to as evolution in materia. The results show that by adding the programmable reservoir layer, reservoir computing in materia can significantly outperform the original evolution in materia implementation. This suggests the RC framework offers improved performance, even across non-temporal tasks, when combined with the evolution in materia technique.",

author = "Matthew Dale and Susan Stepney and Miller, {Julian Francis} and Trefzer, {Martin Albrecht}",

year = "2017",

month = jul,

day = "3",

doi = "10.1109/IJCNN.2017.7966119",

language = "English",

pages = "2178--2185",

booktitle = "International Joint Conference on Neural Networks (IJCNN)",

publisher = "IEEE",

address = "United States",

}

TY - GEN

T1 - Reservoir computing in materio: A computational framework for in materio computing

AU - Dale, Matthew

AU - Stepney, Susan

AU - Miller, Julian Francis

AU - Trefzer, Martin Albrecht

PY - 2017/7/3

Y1 - 2017/7/3

N2 - The Reservoir Computing (RC) framework is said to have the potential to transfer onto any input-driven dynamical system, provided two properties are present: (i) a fading memory, and (ii) input separability. A typical reservoir consists of a fixed network of recurrently connected processing units; however recent hardware implementations have shown reservoirs are not ultimately bound by this architecture. Previously, we have demonstrated how the RC framework can be applied to randomly-formed carbon nanotube composites to solve computational tasks. Here, we apply the RC framework to an evolvable substrate and compare performance to an already established in materia training technique, referred to as evolution in materia. The results show that by adding the programmable reservoir layer, reservoir computing in materia can significantly outperform the original evolution in materia implementation. This suggests the RC framework offers improved performance, even across non-temporal tasks, when combined with the evolution in materia technique.

AB - The Reservoir Computing (RC) framework is said to have the potential to transfer onto any input-driven dynamical system, provided two properties are present: (i) a fading memory, and (ii) input separability. A typical reservoir consists of a fixed network of recurrently connected processing units; however recent hardware implementations have shown reservoirs are not ultimately bound by this architecture. Previously, we have demonstrated how the RC framework can be applied to randomly-formed carbon nanotube composites to solve computational tasks. Here, we apply the RC framework to an evolvable substrate and compare performance to an already established in materia training technique, referred to as evolution in materia. The results show that by adding the programmable reservoir layer, reservoir computing in materia can significantly outperform the original evolution in materia implementation. This suggests the RC framework offers improved performance, even across non-temporal tasks, when combined with the evolution in materia technique.

U2 - 10.1109/IJCNN.2017.7966119

DO - 10.1109/IJCNN.2017.7966119

M3 - Conference contribution

SP - 2178

EP - 2185

BT - International Joint Conference on Neural Networks (IJCNN)

PB - IEEE

ER -