Investigating Echo-State Networks Dynamics by Means of Recurrence Analysis
Bianchi, FM; Livi, L; Alippi, C
Date: 2 December 2016
Journal
IEEE Transactions on Neural Networks and Learning Systems
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Publisher DOI
Related links
Abstract
In this paper, we elaborate over the well-known interpretability issue in echo-state networks (ESNs). The idea is to investigate the dynamics of reservoir neurons with time-series analysis techniques developed in complex systems research. Notably, we analyze time series of neuron activations with recurrence plots (RPs) and recurrence ...
In this paper, we elaborate over the well-known interpretability issue in echo-state networks (ESNs). The idea is to investigate the dynamics of reservoir neurons with time-series analysis techniques developed in complex systems research. Notably, we analyze time series of neuron activations with recurrence plots (RPs) and recurrence quantification analysis (RQA), which permit to visualize and characterize high-dimensional dynamical systems. We show that this approach is useful in a number of ways. First, the 2-D representation offered by RPs provides a visualization of the high-dimensional reservoir dynamics. Our results suggest that, if the network is stable, reservoir and input generate similar line patterns in the respective RPs. Conversely, as the ESN becomes unstable, the patterns in the RP of the reservoir change. As a second result, we show that an RQA measure, called Lmax, is highly correlated with the well-established maximal local Lyapunov exponent. This suggests that complexity measures based on RP diagonal lines distribution can quantify network stability. Finally, our analysis shows that all RQA measures fluctuate on the proximity of the so-called edge of stability, where an ESN typically achieves maximum computational capability. We leverage on this property to determine the edge of stability and show that our criterion is more accurate than two well-known counterparts, both based on the Jacobian matrix of the reservoir. Therefore, we claim that RPs and RQA-based analyses are valuable tools to design an ESN, given a specific problem.
Computer Science
Faculty of Environment, Science and Economy
Item views 0
Full item downloads 0
Related items
Showing items related by title, author, creator and subject.
-
Diet analysis of two data deficient stingray species, the southern stingray (Hypanus americanus) and the Caribbean whiptail ray (Styracura schmardae), with methodological insights into the use of stomach content analysis and stable isotope analysis
Wrigglesworth, E (University of Exeter Biological Sciences, 18 March 2019)The study of animal diet has long been a fundamental area of biological sciences and has developed significantly over the centuries. An understanding of animal diet goes beyond species-specific biology providing insights ... -
A Comprehensive Analysis of Policy Diffusion: Regulatory Impact Analysis in EU and OECD Member States
De Francesco, Fabrizio (University of Exeter Politics, 23 July 2010)Among the tools available to enhance the rationality of policy formulation, Regulatory Impact Analysis (RIA) has captured the attention of many scholars for its potential to enhance the accountability and transparency of ... -
An International Relations Analysis of Citizenship and Intercultural Dialogue among Minority Youth in Berlin and London: A Levels-of-Analysis Approach
Difato, Christine Anne (University of Exeter Exeter Centre of Ethno-Political Studies, 30 September 2012)Corresponding to the rising threat of terrorism and heightened security concerns throughout Europe, potential parallel communities have come under increased scrutiny and pressure to engage and integrate in the politics and ...