Enhancing neural-network performance via assortativity

Journal Article · Tue Mar 15 00:00:00 EDT 2011 · Physical Review. E, Statistical, Nonlinear, and Soft Matter Physics (Print)

DOI:https://doi.org/10.1103/PHYSREVE.83.036114· OSTI ID:21560072

Franciscis, Sebastiano de; Johnson, Samuel; Torres, Joaquin J ^[1]

---

1. Departamento de Electromagnetismo y Fisica de la Materia, and Institute Carlos I for Theoretical and Computational Physics, and Facultad de Ciencias, University of Granada, E-18071 Granada (Spain)

The performance of attractor neural networks has been shown to depend crucially on the heterogeneity of the underlying topology. We take this analysis a step further by examining the effect of degree-degree correlations - assortativity - on neural-network behavior. We make use of a method recently put forward for studying correlated networks and dynamics thereon, both analytically and computationally, which is independent of how the topology may have evolved. We show how the robustness to noise is greatly enhanced in assortative (positively correlated) neural networks, especially if it is the hub neurons that store the information.

• OSTI ID: 21560072
• Journal Information: Physical Review. E, Statistical, Nonlinear, and Soft Matter Physics (Print), Vol. 83, Issue 3; Other Information: DOI: 10.1103/PhysRevE.83.036114; (c) 2011 American Institute of Physics; ISSN 1539-3755
• Country of Publication: United States
• Language: English

Related Subjects

71 CLASSICAL AND QUANTUM MECHANICS
GENERAL PHYSICS
46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND TECHNOLOGY
ANALYTICAL SOLUTION
ATTRACTORS
CORRELATIONS
INFORMATION
NERVE CELLS
NEURAL NETWORKS
NUMERICAL SOLUTION
PERFORMANCE
TOPOLOGY
ANIMAL CELLS
MATHEMATICAL SOLUTIONS
MATHEMATICS
SOMATIC CELLS