Anybody interested in genetic/evolutionary algorithms, general heuristics as well as neural networks and fuzzy logic?

hi, my thesis was a ga based research and analysis of two end fire antennas.
whats up i am always interested,

My final year undergraduate project was on artificial neural networks.
Though I still consider myself a novice in the field, but I was able to simulate a
backpropagation neural network that shows you can compress any digital data to half
it's original size.

It was then implemented in the java programming language, but I no longer code in java now,
I was thinking of reimplementing it in C# or VB.NET.

Alright, what was the thesis about? I have worked on the N-Queen problem, Traffic flow optimization, knapsack resource allocation etc.
If you have any questions, don't hesitate to ask.

IG,
I would suggest you use MATLAB instead, it comes with a neural network toolbox.
That will enable you focus on the essentials and speed up your project execution.

IG:

My final year undergraduate project was on artificial neural networks.
Though I still consider myself a novice in the field, but I was able to simulate a
backpropagation neural network that shows you can compress any digital data to half
it's original size.

It was then implemented in the java programming language, but I no longer code in java now,
I was thinking of reimplementing it in C# or VB.NET.

IG, i'm going to resume to final yr this later this yr. could u pls send a copy of ur project to my mailbox for review. i may like to use it as my final yr project too. its doc_lanre@yahoo.com

Larrydee:

IG, i'm going to resume to final yr this later this yr. could u please send a copy of your project to my mailbox for review. i may like to use it as my final yr project too. its doc_lanre@yahoo.com

Yeah Review. Mr. Supervisor.

Does the Word Plagiarism Mean Anything To You. Im Sure Uve Heard That B4.

i worked on a PHD dissertation of a well known but late Professor of petroleum Engineering from Ife, my project sought to use ANN to substitute conventional statistical and manual simulation techniques such as AVRIMA ,ARIMA, LMT used by the prof in predicting the Perfomance of Insitu Combustion process in Heavy oil recovery.
The project won the most Original Work in my Department , and ranked 5th best project in the school. Its on display in the Library right now. In world publications on petroleum, the subject and method as applied to Insitu combustion is yet to be achieved by any. so my project is first and original. But i must i need deeper knowledge of the Process practically to further on the work.

@MOG, I learned that MATLAB can do that. But I didn't have the time to learn MATLAB but I already know Java then and my supervisor insist on doing everything from scratch.

Anyway in a nutshell, I implemented a 3 layered backpropagation neural network.
The input and output layers have eight neurons each while the middle layer has four neurons.

After training the neural network to a learned state. We were able to input 8-bit binary data, one bit in each neuron and recover the exact data in the output layer.

This means that we are essentially squeezing an 8-bit data into a 4-bit channel ( the middle layer) and recovering it.
In other words compressing it to half it's size and decompressing it.

If you guys are really interested in it, I will serch for the soft copy of the thesis( I sold it on my old desktop) and upload it here.

But don't forget I said that I still consider myself a novice in the field.

@Pain, in the end parts of my thesis, I suggested ways of improving my work. So you can have something to work on without necessarily plagiarising my work.

Hmmn!! Neural networks and G.A? Nice combination. I used ANN for my project thesis to forecast Stock price and Traffic Vehicular movement. I also used a fuzzifier to decode the output from the A.N.N. I was hoping I could implement a G.A. for the network cause of local optima trapping.

@IG, with regards to neural networks, I would recommend MATLAB's neural network toolbox for anyone with a critical project.

It comes with good documentation in pdf format. With the toolbox, learning is fast and re-work is reduced (you don't have to build everything from scratch).

For newbies to genetic or evolutionary programming, it a heuristic approach that especially usefully in solving problems with high computational complexity e.g. nondeterministic polynomial time hard (NP hard) problems and in many other types of problems - time varying, highly constrained etc.

The basic concept is to maintain (in computer memory!) a population of potential solutions and perform operations like crossover, mutation, variation and of course selection over a number of generations; all the while evaluating each sucessive population for the overall best. There is not predertermined routine for any problem. It is left to your imagination and ingenuity. Your skill can develop if you study relevant literature and practise.

hey guys
could you exolain what geneteic algorithm and neural networks means
in layman's language

Am new here in nairaland. I will like to make programming friends from across Nigeria.
I am an undergraduate of Computer Science at the Federal University of Technology, Akure, Nigeria. I am also a Sun Certified Java Programmer and also an Orcale Certified Profesional.

pls i'd like to have a look at ur project ucan send it to my box tobigbeja@yahoo.co.uk. thnx

uhm, Send your project for tobbyjohn to have a look at.
A smart man does not need more than "look at" the work/project of a stupid man to finish it up for him.