Figure 4
The development of HNSM and its results. (a) Evolution of HNSM in different task complexity. (b) Illustration of HNSM, including data flow, neuron definition, relationship of matrices and the whole structure. (c) The variation curve of error transfer during training. (d) The weights of T-T, T-S, S–S, S-T matrices. (e) State-maintenance and disturbance rejection. (f) Showcase of a complex state transition sequence. Software: Microsoft Visio 2019 MSO (16.0.10730.20102) 64-bit. https://www.microsoft.com/en-us/microsoft-365/visio/flowchart-software MATLAB R2017a (9.2.0.538062) 64-bit (win64) https://ww2.mathworks.cn/products/matlab.html.
