Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Strategies & Market Trends : Neural Nets - A tool for the 90's -- Ignore unavailable to you. Want to Upgrade?

To: current trend who wrote (24)	12/30/1997 11:34:00 AM
From: Optim	Respond to of 871

>Since input selection, management, and massage are so critical >to the ultimate neural net output---would you speak to your >experiences and offer advice as to how to best aproach this task? This is something that is still a work in progress. I have only been using these NN packages for a few months, so there is still much to learn. However, by using a GA to evolve good selections of inputs you can shorten the development cycle. This is something I haven't had time to play with much over the holidays, but I know will pay great dividends if I dedicate the time to it. I work as a contract systems engineer so I don't get much holiday time, and having only had the GeneHunter GA package for about 2 weeks now, I haven't really had the opportunity to put it through its paces. I will keep the thread posted though... >Over optimization is the first deadly sin of TA... not so with NN? True. A net can suffer the same problems. You want to use the minimum number of inputs possible, while still maintaining a robust set that will work in all types of markets. Again I have yet to find a good combination, so I really can't give you much advice yet. >Are the nets finding non linear relationships that we can not >see or just using computational power to discover subtilities >about the edges that produce their effectiveness? This is exactly what an ANN does. It finds subtle, non-linear relationships amongst your inputs and desired output, and then produces a prediction or guess as to what may occur based on current inputs. You don't really have to understand what is going on inside the net, but it may help you find better inputs by at least understanding how it works at a conceptual level. Optim