Reservoir Computing for Scalable and Energy-Efficient AI: Theory, Dynamics, and Implementations
Special Session of the International Joint Conference on Neural Networks 2026
21 - 26 June 2026, Maastricht, the Netherlands
21 - 26 June 2026, Maastricht, the Netherlands
Paper submission deadline: January, 31th 2026 (AoE)
Decision notification: March, 15th 2026
Reservoir Computing (RC) has established itself as an efficient and versatile approach for training and designing recurrent neural networks (RNNs), offering powerful capabilities in processing temporal and spatio-temporal data without the heavy computational demands of conventional deep learning models. As modern artificial intelligence increasingly moves toward large-scale, low-latency, and energy-efficient computation, the relevance of RC extends to a broad range of applications, including signal processing, robotics, neuroscience, and edge computing. The growing interest in hardware-aware and resource-constrained AI also reinforces the importance of RC as a framework for neuromorphic and embedded intelligence. However, to fully meet the challenges posed by scalability, robustness, and efficiency, RC requires new advances in theory, model design, and implementation strategies.
The goal of this special session is to promote progress in Reservoir Computing by addressing its theoretical, algorithmic, and hardware dimensions. The first objective is to deepen the theoretical understanding of the dynamical principles that govern stability, adaptability, and generalization in reservoir systems, particularly in the context of large-scale and heterogeneous data. The second objective is to foster the development of hybrid architectures that combine RC with modern deep learning approaches (such as convolutional, graph-based, or hierarchical models) to extend its reach to complex spatio-temporal domains. The third goal is to explore efficient and energy-aware hardware realizations of RC, leveraging advances in neuromorphic, photonic, and spintronic technologies for real-time, low-power computation.
We invite researchers to submit papers on all aspects of RC research, targeting contributions on theory, models, and applications.
A list of topics relevant to this session includes, but is not limited to, the following:
Theoretical foundations of Reservoir Computing, including stability analysis, expressivity, generalization, and memory–capacity characterizations
Dynamical systems theory for reservoir design, adaptation, and analysis
Novel reservoir architectures: structured, modular, deep, hierarchical, multiscale, or hybrid with modern deep learning paradigms
Learning algorithms for reservoir systems, including gradient-based, constrained, biologically inspired, or unsupervised/self-supervised approaches
Training and adaptation of recurrent dynamics: online learning, continual learning, meta-learning, and federated settings in RC
Reservoir Computing for high-dimensional spatio-temporal data, including images, video, physical processes, and graph-structured domains
Hardware-oriented RC: neuromorphic, photonic, spintronic, analog, and mixed-signal implementations; co-design of models and substrates
Energy-efficient and resource-aware RC for embedded, edge, or on-device intelligence
Physical reservoirs and in-materio computing for machine learning
Reservoir Computing in computational neuroscience and brain-inspired modeling
Robustness, scalability, and reliability of reservoir systems in real-world environments
Applications of Reservoir Computing across scientific, industrial, and societal domains, including robotics, system identification, forecasting, signal processing, geoscience, and biomedical data analysis
As in recent editions, the review process for IJCNN 2026 will be double-blind.
Page limit for full papers: up to 6 pages; notice that a maximum of two extra content pages per full paper is allowed (i.e, up to 8 pages), at an additional charge per extra page as specified in the registration page of the conference. Please note that ORCID will be required for each author of the paper.
Full authors' instructions can be found at the following link: https://attend.ieee.org/wcci-2026/information-for-authors/
Further information on the submission process can be found at the following link: https://attend.ieee.org/wcci-2026/submissions/
Link to submission system: https://ssl.linklings.net/conferences/WCCI/
Note that to submit to this special session, you need to create a new submission (Make a New Submission tab) and select the track IJCNN SS27 Reservoir Computing for Scalable and Energy-Efficient AI: Theory, Dynamics, and Implementations.
Benjamin Paassen (Bielefeld University, Germany), Gouhei Tanaka (Nagoya Institute of Technology, Japan), Andrea Ceni (University of Pisa, Italy), Claudio Gallicchio (University of Pisa, Italy),