Abstract:Graph reservoir computing (GraphRC) gains increasing attention by virtue of its high training efficiency. However, since GraphRC is developed without knowledge of its internal mechanism, it cannot be fully trusted to deploy in practice. Although there are some existing approaches that can be extended to interpret GraphRC, the specific role played by each neuron (i.e., reservoir node) of GraphRC is far less explored. To address this issue, the latent short-term memory property of each reservoir node of GraphRC is qualitatively characterized to unravel its role in predicting the graph signal, thereby enabling an interpretable GraphRC. Specifically, we first deduce the equivalence between the GraphRC and conventional reservoir computing (RC). Then, the underlying memory properties of the GraphRC and its reservoir nodes can be characterized in theory by the multisource reachability among the reservoir nodes in the transformed RC. Moreover, the distinct temporal patterns hidden in reservoir nodes are identified, and then, an attention mechanism based on the identified temporal patterns is deployed in the GraphRC to improve its performance. In addition, the effectiveness of the interpretability for GraphRC and improved GraphRC is verified on the Lorenz-96 spatiotemporal dynamical system. The experimental results of the Lorenz-96 spatiotemporal chaotic system and three real-world traffic datasets demonstrate that the improved GraphRC is superior to original GraphRC and can achieve prediction performance comparable to the state-of-the-art baseline models, but with much less training cost.

Explaining Memristive Reservoir Computing Through Evolving Feature Attribution.

Efficient Neuromorphic Reservoir Computing Using Optoelectronic Memristors for Multivariate Time Series Classification.

Evolving Memristive Reservoir.

Dynamic Memristor-based Reservoir Computing for High-Efficiency Spatiotemporal Signal Processing

Analysis and Fully Memristor-based Reservoir Computing for Temporal Data Classification

Application of Reservoir Computing Based on a 2D Hyperchaotic Discrete Memristive Map in Efficient Temporal Signal Processing.

A Feature-Recombinant Asynchronous Deep Reservoir Computing for Modeling Time Series Data

Interpretable Memristive LSTM Network Design for Probabilistic Residential Load Forecasting

Enhancement of Multi-Timescale Reservoir Computing Based on Homogeneous Hardware Platform with Controllable Ionic Dynamics

Self-organizing Reservoir Computing Based on Spiking-Timing Dependent Plasticity and Intrinsic Plasticity Mechanisms

Reservoir Computing with Diverse Timescales for Prediction of Multiscale Dynamics

Reservoir computing dissection and visualization based on directed network embedding

An Intelligent Interconnected Network with Multiple Reservoir Computing

Self-Evolutionary Reservoir Computer Based on Kuramoto Model

Reservoir computing using dynamic memristors for temporal information processing

A Multi-Terminal Ion-Controlled Transistor with Multifunctionality and Wide Temporal Dynamics for Reservoir Computing

Learning reservoir dynamics with temporal self-modulation

Reservoir Computing Decoupling Memory-Nonlinearity Trade-Off.

Deep Reservoir Computing Based on Self-Rectifying Memristor Synapse for Time Series Prediction

Adaptive Memory-Enhanced Time Delay Reservoir and Its Memristive Implementation

Interpretable Graph Reservoir Computing With the Temporal Pattern Attention