Dynamic adaptive learning environment construction system and method of morphological neural network

Abstract

The invention discloses a dynamic adaptive learning environment construction system and method for a morphological neural network, and the system comprises a multi-modal data cooperative processing module which collects corresponding signals through a pressure sensor, an eye movement sensor and an electroencephalogram sensor, and generates a high-reliability fusion instruction stream through a cross-modal joint verification mechanism after decomposition and denoising; the neuromorphic dynamic modeling module converts the data into a biological neuron type pulse sequence through a pulse graph neural network, and constructs a dynamic knowledge topology to generate a spatial-temporal characteristic graph; the hybrid neural network training and optimization module dynamically adjusts synaptic weights based on corresponding strategies to optimize a network structure; and the cognitive state adaptive rendering module maps the spatial-temporal characteristics to a low-dimensional hidden space through a neural rendering compression engine, and generates an adaptive interface in real time. According to the scheme, each original signal is subjected to feature extraction and processing, and the features of different modal data are subjected to deep fusion to generate the fusion instruction stream, so that accurate mapping of neural signals and external behaviors is realized.

Images