The role of structure and complexity on Reservoir Computing quality

TY - CONF

T1 - The role of structure and complexity on Reservoir Computing quality

AU - Dale, Matthew

AU - Dewhirst, Jack Daniel

AU - O'Keefe, Simon Edward Marius

AU - Sebald, Angelika Anne-Marie

AU - Stepney, Susan

AU - Trefzer, Martin Albrecht

PY - 2019

Y1 - 2019

N2 - We explore the effect of structure and connection complexity on the dynamical behaviour of Reservoir Computers (RC). At present, considerable effort is taken to design and hand-craft physical reservoir computers. Both structure and physical complexity are often pivotal to task performance, however, assessing their overall importance is challenging. Using a recently proposed framework, we evaluate and compare the dynamical freedom (referring to quality) of neural network structures, as an analogy for physical systems. The results quantify how structure affects the range of behaviours exhibited by these networks. It highlights that high quality reached by more complex structures is often also achievable in simpler structures with greater network size. Alternatively, quality is often improved in smaller networks by adding greater connection complexity. This work demonstrates the benefits of using abstract behaviour representation, rather than evaluation through benchmark tasks, to assess the quality of computing substrates, as the latter typically has biases, and often provides little insight into the complete computing quality of physical systems.

AB - We explore the effect of structure and connection complexity on the dynamical behaviour of Reservoir Computers (RC). At present, considerable effort is taken to design and hand-craft physical reservoir computers. Both structure and physical complexity are often pivotal to task performance, however, assessing their overall importance is challenging. Using a recently proposed framework, we evaluate and compare the dynamical freedom (referring to quality) of neural network structures, as an analogy for physical systems. The results quantify how structure affects the range of behaviours exhibited by these networks. It highlights that high quality reached by more complex structures is often also achievable in simpler structures with greater network size. Alternatively, quality is often improved in smaller networks by adding greater connection complexity. This work demonstrates the benefits of using abstract behaviour representation, rather than evaluation through benchmark tasks, to assess the quality of computing substrates, as the latter typically has biases, and often provides little insight into the complete computing quality of physical systems.

KW - Complexity

KW - Echo state networks

KW - Reservoir computing

KW - Structure

KW - Unconventional computing

UR - http://www.scopus.com/inward/record.url?scp=85066806136&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-19311-9_6

DO - 10.1007/978-3-030-19311-9_6

M3 - Paper

SP - 52

EP - 64

T2 - UCNC 2019, Tokyo, Japan, June 2019

Y2 - 3 June 2019 through 7 June 2019

ER -