Artificial Neural Networks URI BME Aleksey Gladkov Introduction

Artificial Neural Networks URI BME Aleksey Gladkov

Introduction An artificial neural network is a mathematical or computational model that approximates the structure or function of biological neural networks. (Pictured: Alvin)

Artificial Neurons are modeled using a function which responds to various weighted inputs, and is capable of adjusting the weights over time as it “learns”.

Uses Artificial neural networks can be used to model complex relationships between many variables, as well as being able to spot patterns in a large quantities of data.

Drawbacks A major problem with artificial neural networks is the amount of work that must be put into the “learning” step of development. Another issue is the amount of processing and storage required to maintain such a network with the currently available technology.

Applications -Facial Recognition -Manufacturing Process and Quality Control -Handwriting and Speech Recognition -Spam Filtering -Gene Recognition -Many Kinds of Forecasting -Physical System Modeling

Memresistors A memresistor is a two terminal device which changes its resistive properties depending on the direction of current passing through the device. The most important feature of this substance is the ability to retain its resistive properties even when there is no current present. Memristor theory was formulated and named by Leon Chua in a 1961 paper. In 2008 HP Labs announced the development of a switching memristor based on a thin film of titanium dioxide.

What This Means Solid-state memristors can be combined into devices called crossbar latches, which could replace transistors in future computers, taking up a much smaller area. HP prototyped a crossbar latch memory using the devices that can fit 100 gigabits in a square centimeter.

Fuzzy Logic Unlike Binary Logic, which has exact values for TRUE and FALSE (0, 1 respectively), Fuzzy Logic has been extended to handle the concept of partial truth, where the truth value may range between completely true and completely false.

What This Means Fuzzy logic is a lot more realistic than binary logic, so it can make the artificial neuron algorithm more realistic and lifelike.

Current Research At this time, researchers are attempting to evaluate the effects of neuromodulators on natural networks in order to better understand their functions, which will, hopefully, give us some insight on better simulating them with artificial ones.

The Future of Artificial Neural Networks With further development of memresistor technology and miniaturization, and liberal application of fuzzy logic, artificial neural networks can accurately mimic the functions of natural ones. After that is only a matter of time before computers that learn as the go.

The End?

References Alyuda. 24 Sep. 2011. <http: //www. alyuda. com/products/forecaster/neural-network-applications. htm>. Gershenson, Carlos. "Artificial Neural Networks for Beginners. " Cornell University. 24 Sep. 2011. <http: //arxiv. org/ftp/cs/papers/0308031. pdf>. Miller, Michael J. "Memristors: A Flash Competitor that Works Like Brain Synapses. " 2010. Forward Thinking. 26 Sep. 2011. <http: //forwardthinking. pcmag. com/chips/282616 -memristors-a-flash-competitor-that-works-like-brainsynapses#fbid=x 6 JCe. Ar. NPkc>. Regine. "BRAINWAVE: Common Senses. " 26 Sep. 2011. <http: //www. we-make-money-notart. com/archives/2008/03/brainwave-common-senses. php>. Stergiou, Christos and Dimitrios Siganos. "NEURAL NETWORKS. " Imperial College London. 24 Sep. 2011. <http: //www. doc. ic. ac. uk/~nd/surprise_96/journal/vol 4/cs 11/report. html>. Wikipedia. 24 Sep. 2011. <http: //en. wikipedia. org/wiki/Artificial_neural_network>. All Images Courtesy of Google Images