Reservoir Computing: Trends and Future Prospects

Deep learning and machine learning techniques are at the forefront of Artificial Intelligent (AI) in various fields. One of the important computing methods used to overcome the training difficulty in deep learning and machine learning techniques is Reservoir Computing (RC). Even after years of discovery, reservoir computing remains applicable for all the advanced technologies involving Recurrent Neural Networks (RNN). The RC approach started with Echo State Networks (ESN) and Liquid State Machines (LSM). But now the application of RC has extended to all the other fields of research such as standard machine learning, input-output mapping, etc. The recent applications of RC are in speech recognition with natural language processing, handwriting recognition in certain software and companies and robot motor controlling by tracking and managing non-linear plants of ESN. Adding to that, RC is also employed in financial forecasting, the bio-medical sector and healthcare. Different types of RC such as echo state network, non-linear transient computation and deep reservoir computing are employed in different sectors based on the usage.

Between physical reservoir computing and RNN-based reservoir computing, the former is used for natural science-related researches such as biology, chemistry and physics. The latter is majorly used in easing the usage of deep learning. As RC has the knowledge and data to predict the placement of a point with time, it is now being used in time series prediction, ESN predictors, etc. In the field of biology, RC is used in the detection of diseases, cognitive prediction, neuroscience and medication. The recent trends in RC include the application in financial markets. The changing world scenario needs micro and macro-economic prediction for an acceleration of economy and business. Thus, RC with its capability to model complex data sets can be used in financial modelling as well. As RC is suited for temporal data processing, it can be applied in electronic, photonic and mechanical modelling and research. The context reverberation systems can be used in high-dimensional dynamical systems. The main advancement in RC is quantum reservoir computing that enables quantum mechanical interaction. The other recent advancements in RC are reservoir memory machines, pyramidal state echo networks, simplified deep reservoir architectures, self-organised dynamic attractors in recurrent neural networks, self-organised echo state networks and human action recognition.  Hardware RC implementations with low-cost computation can be used for efficient training. Additionally, RC can also be used in memory augmented neural networks. One main advantage of RC is its universality. The application of RC in all the fields and related advancements make it more reliable. With the increased application, the non-compromising accuracy in RC is another added benefit. This issue tends to research the current trends and future prospects of reservoir computing.

List of Topics (include, but not limited to the following):

  • Modern deep networks and its application in reservoir computing
  • Convolutional filtering operation with reservoir computing
  • Training mechanism in deep learning enhanced with reservoir computing
  • Exotic hardware implementation systems with reservoir computing
  • Mathematical foundations for the accurate usage of reservoir computing
  • Theory of dynamical systems and its application in reservoir computing
  • Presence of external driving inputs in reservoir dynamics
  • Deep echo state networks in reservoir computing
  • Dynamical recurrent models based on reservoir computing in deep learning
  • Untrained neural architecture for graphs using reservoir computing


Important Dates:

  • Manuscript Submissions Deadline: 15.04.2022
  • Notification to Authors: 05.07.2022
  • Deadline for revision submissions: 10.10.2022       
  • Notification of final decisions: 30.12.2022
  • Tentative Publication: As per the journal’s policy 

Guest Editor Details:

Dr. Ahmed A. Abd El-Latif, Associate Professor, Department of Mathematics and Computer Science, Faculty of Science, Menoufia University, Shebin El-Koom, Egypt.

Dr. Edmond Shu-lim Ho, Senior Lecturer, Department of Computer and Information Sciences, Northumbria University, Newcastle upon Tyne, United Kingdom. 

Dr. Jialiang Peng, Associate Professor, School of Data Science and Technology, Heilongjiang University, Harbin, China

Download full details hereReservoir Computing: Trends and Future Prospects