-2

I am looking to create an artificial neural network that is able to recognize simple stock market patterns that occur over a few days and then I want to be able to graphically display the output. For example if a user typed in the name of a company then clicked "GO" then the program would see if there were any patterns in the last 10 days and output "pattern found".

I am just having some trouble getting started. I am comfortable programming in Java but I am totally new to neural networks. I was considering writing the interface/output part as a webpage and then maybe writing the neural network in JavaScript.

I would prefer to write the neural network in Java but im not sure how to write graphics etc in Java.

Can anybody give me some guidance as to how I would write a pattern recognition neural network for the stock market? could I use the prices as inputs and then just a boolean 1 or 0 as the output if the pattern is found or not?

Thanks

James
  • 111
  • 2
  • 4
  • 15
  • Have a look at HTMs: in particular the work of Numenta – David Lavender Feb 18 '13 at 15:39
  • thank you for the suggestion :) I have looked into HTM but it seems even more complex than neural networks. I only want it to recognize 1 or 2 patterns and so want to keep things as simple as possible. – James Feb 18 '13 at 15:44

1 Answers1

0

I guess Java is a good choice!

Indeed you could use JSP technology to render your web pages and display all graphic stuff with a javascript library (e.g. using the canva tag). And for neural networks you can use the well known weka library which enables the use of dozens of machine-learning algorithms (see here).

Davd33
  • 20
  • 3