ABSTRACT
Biological neurons communicate using electro-chemical action potentials, they also possess a variety of slower, chemical regulatory processes that analyze the statistical properties of information flow. The execution of higher functions such as learning, memory or pattern recognition requires a collective orchestration of how the neural network is interconnected. If digital computers possess a large amount of memory and complex instructions stored in a single memory stack, a neural network must distribute its instructions in neurons and its memory in the networking between these neurons. The principle states that the goal of a layered perceptron neural network’s learning procedure is to maximize the mutual information between its output and inputs. A combination of intrinsic plasticity and synaptic plasticity in individual neurons can organize a network so that it can solve principal component analysis and a classical formulation of independent component analysis. In biological networks, the spike-timing dependent plasticity mechanism is governed by biomolecular protein transmitters and receivers.
