The Simplest Artificial Neuron

15 Apr 2016

The human brain is a complicated system that is responsible for every single human accomplishment and endeavor. It is the best example we currently have of intelligence. This is the kind of intelligence that researchers are currently trying to emulate. If we are trying to understand the brain at it's most basic level, We can started by learning about neurons. Neurons are the underlying cause of all human congnition and behavior. Neurons are also not unique to humans, most animals have neurons. These neurons communicate to each other with electrical signals and chemical processes. The neurons in the nervous system communicate through axon terminals and dendrites. This means every neuron has an input and an output.

An artificial neuron is also structured in the same way (It has an input and an output). An artificial neuron has weight vector which determines how important an input is for the kind of output that we want from the neuron. The artificial neuron also has an activation function that gives us the actual output of the neuron. We take the dot product of the input vector and the weight vector to get the input that we need for our activation function. Here is a quick summary of the Dot Product. The activation function that is commonly used is The Sigmoid Function. Here is a quick summary of the Sigmoid Function.

Summary: A simple Artificial Neuron is a Sigmoid Function whose input is the dot product of the weight vector and the input vector.

Discuss on Github

Reference:

Nielsen, Michael. "CHAPTER 1." Neural Networks and Deep Learning. VisionSmarts,Ersatz,G Squared Capital,TinEye, 22 Jan. 2016. Web. 15 Apr. 2016.

To the extent possible under law, Victory Omole has waived all copyright and related or neighboring rights to this work by licensing it under a Public Domain License.