Reservoir Self Organizing Map
Hiroshi Dozono, Ryuhei Matsuo, Koki Yoshioka · 2022 Joint 12th International Conference on Soft Computing and Intelligent Systems and 23rd International Symposium on Advanced Intelligent Systems (SCIS&ISIS) · 2022
As IoT devices become more widespread, the load on cloud servers that process the time-series data acquired from IoT devices are expected to increase. For this reason, edge computing, in which data processing should be performed by IoT terminals and servers installed nearby, is attracting attention. It is difficult to use high-performance computers that can handle large amounts of computation on edge computing. As a solution to this problem, Reservoir Computing, a kind of recurrent neural network with low computational cost, was proposed. The purpose of this research is to improve the prediction accuracy of time series data by combining Reservoir Computing with SOM and changing the method for updating the weight vectors of output.