International Journal of Enhanced Research in Science, Technology & Engineering ISSN: 2319-7463, Vol. 5 Issue 11, November-2016 Page | 1 An Innovative Architecture to Model Human Brain as Cognitive Computation System Alaa AL-MISSTTAF, Hussein CHIBLE, Rami Tawil, Ammar FATTAH MECRL Lab, PhD School for Science and Technology, Lebanese University, Beirut, Lebanon Information Technology Department, ISCC, Baghdad, Iraq ABSTRACT This paper presents novel approach to model the human brain functionality as a cognitive computation system. Here the brain appears as two different levels: the sensor level (i.e., object level) and the concept level (i.e., ontological level). Each level has a different stimulation pattern. Concept level is dominant over the sensor level due to the hierarchal structure combining those levels. Using a new Perceptron model is important in achieving the intended goals that can be summarized in: a) Ability to preserve the input’s importance. b) Ability to perform both temporal and spatial neuronal summation. c) Ability to dynamically change its structure by undergoing through rewiring condition when recognizing a new object. d) Ability for continuous learning and gaining experience with frequent practicing. The new architecture makes the brain seems as a cognitive system in which the basic unit of function (i.e. neurons) interoperability is best described using linear algebra principals. The system is examined by using the well known Iris Flowers dataset. Keywords: Cognition, Experience, Learning, Neural Coding, Rewiring. I. INTRODUCTION During the last century, humans aspire to model the machine that mimics their abilities of thinking, learning and decision- making. These attempts are based on its basic understanding of the physiological functions. The development of the biological sciences together with the development of the imaging devices made the process of studying of human thinking sites easier [1, 2]. Artificial neural networks are considered as computational techniques that emulate how the human brain does its basic functions. The human brain is composed of an enormous number of nerve cells which are considered as the basic block unit of construction and function. Each of these cells are connected to many other nerve cells composing what is biologically known as neural networks and they create a very complex pattern of electrical signal transmission. The signals are transmitted among the neurons through special connecting areas called “synapses” [3]. Researchers started to study these biological facts and begun their attempts to model the basic nervous unit “neuron”. The first attempt to model a neuron was carried out in 1943 by McCulloch and Pitt who first proposed a computational neuron [4, 5]. Next, in 1958 Rosenblatt proposed the first neural network that is known as Perceptron [5, 6]. All the neural networks from their invention till now share common characteristics. They have the same building block unit “neuron” and the interconnection between these blocks units. The most commonly known neuron is the Perceptron modelled by McCulloch [4]. Since the basic idea was inspired from human biology and all the researches in neural network domain depend on the same block unit, the current studies show that there is a big demand to introduce some structural changes to the basic architecture of the Perceptron. The modifications should be done in a way that makes it converge more toward simulating the biological nerve cell. The following sections involve a brief introductive explanation about the subject of this article. II. PERCEPTRON AND ARTIFICIAL NEURAL NETWORK Before talking about Perceptron, we have to take an idea about the biological neuron. It is a special type of cells, composes the nervous system and the brain of humans and other developed species. Neurons have a remarkable property of electrical excitation [7, 8]. A typical neuron is divided into three main parts:  Cell body “Soma”.  Dendrites.  Axons.