Improving Regression in Spiking Neural Networks for Oceanographic Data Analysis

Spiking neural networks (SNNs), as the third-generation neural networks, can work under an energy efficient mode. SNNs are different from the second-generation neural networks which consume a lot of energy and power. SNNs are suitable for oceanographic data analysis on the edge devices underwater since the devices have constrained power supply and limited communication bandwidth in underwater environments. Although SNNs have been widely used in classification tasks, SNN-based regression tasks are studied less because SNNs are generally considered to process discrete and sequential spikes. The existing regression model based on the membrane potential of Leaky Integrate-and-Fire (LIF) neuron uses constant settings and this mechanism may not be adaptive and capable of analyzing oceanographic data which are complicated and dynamic. In this paper, we proposed three novel regression models of Adaptive Threshold Adjustment, Heterogeneous Neurons, and Nonlinear Integration to improve the existing LIF-based model. Experimental results on real oceanographic data indicate that the proposed regression models outperform the existing model through qualitative and quantitative analysis. Those SNN regression models could be implemented on edge devices within underwater environments in the future.