The KIV Model
|
The KIV model is a biologically inspired neural network. K-set family includes hierarchy of K models, including K0, KI, KII, KIII, and KIV, see Fig. 1. K-sets represent family of models with increasing complexity. They represent different aspects of the vertebrate brain. K0 consists of a single node with input and output and a nonlinear transfer function. KI is a coupling of either exhibitory or inhibitory K0 units. KII is a double layer of both exhibitory and inhibitory KI units. KIII is three-layer architecture with one or more KIIs connected by feed forward and delayed feed back neural network. KIII model is a working example of the chaotic principles .The KIII model is used for robust pattern recognition. Finally, KIV is the mathematical model of the brain.
KIV model consists of Hippocampus, Cortical and Amygdala. It has the functionality of sensory perception and action selection. KIV is a chaotic dynamic memory which encodes sensory information in the form of aperiodic spatial-temporal oscillations of non-linear processing elements. Figure 2 provides a KIV model of the hemisphere, Fig. 2a and b. It has three major parts, cortex, hippocampus and enthorhinal cortex with the amygdala, following the limbic system, Fig. 2a. Hippocampus is a KIII neural network which models proprioception including navigation functions. The cortex is another KIII neural network which models sensory processing and pattern recognition in various sensory modalities. The amygdala is a KII neural network which is incorporated into the KIV neural network. It is the unit where the activations from the both KIII are taken and decision is made concerning the next action, based on the fusion of the signals from the KIII sets. Amygdala is linked with both KIIIs as shown in the following figure with some weighted matrix. With proper weight selection the KIV neural network can maintain some non convergent chaotic oscillations among all the components of the system. The activations from Hippocampus and Cortical KIII’s are transferred between them through connection weight matrix WA. The activations between Hippocampus and Amygdala are passed through weighted matrix WB. Finally, the activations between Cortex and Amygdala are passed through weight matrix WC.
The weights between the KIII and KII subcomponents provide the manner in which we can model various states of electrical neuronal activity found in a human EEG. |
