An adaptive NeuroMolecular Computing Net model

Neurons in our brain play a decisive role to perform highly information computing. Most of the artificial neural models merely emphasized the inter-neuronal relationship. Ignoring intra-neuronal dynamics is a simplification that might reduces the computational capability of the neurons. In this paper, a bio-inspired NeuroMolecular Computing Net (NMCN) model, which integrates inter-and intra-neuronal information processing so as to capture the biology-like malleability and gradual transformability, was proposed. The model was further applied to medical diagnosis with clinical database of ventilator-dependent patients. Experimental results show that the NMCN model is capable of learning to differentiate data in an autonomous manner. It also achieves a satisfactory result in data differentiation.