Comments on: Hebbian novelty filters for financial analysis http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/ The Peltarion Blog Wed, 07 Jan 2009 10:10:44 +0000 http://wordpress.org/?v=2.7 hourly 1 By: Dirk van de Broek http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/comment-page-1/#comment-907 Dirk van de Broek Tue, 12 Dec 2006 15:33:29 +0000 http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/#comment-907 Hey Thanks for this great read, i have bookmarked your site so i can revisted it Keep up the good work Dirk Hey

Thanks for this great read, i have bookmarked your site so i can revisted it

Keep up the good work

Dirk

]]> By: Mankand Trivedi http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/comment-page-1/#comment-83 Mankand Trivedi Tue, 29 Aug 2006 04:07:05 +0000 http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/#comment-83 Guys, You shoudl check out this site... http://ww.stunnets.com... This site claims to predict stock prices using neural network technology! cheers! Guys,
You shoudl check out this site… http://ww.stunnets.com... This site claims to predict stock prices using neural network technology!

cheers!

]]> By: Pete (Peltarion) http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/comment-page-1/#comment-16 Pete (Peltarion) Sat, 03 Jun 2006 14:50:17 +0000 http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/#comment-16 Sure, instead of connecting the input directly, pipe it through a Gamma Memory component and you'll get the temporal information as well. Alternatively, you can create a recurrent link from the output to a merger where it joins with the signal from a data source. I'd recommend the former though as gamma memories are more flexible in terms of memory depth and resolution than plain feedback loops. Including the temporal dynamics can indeed be useful - the only reason why it wasn't included in the tutorial is because I wanted to keep things simple. Pete Sure, instead of connecting the input directly, pipe it through a Gamma Memory component and you’ll get the temporal information as well.

Alternatively, you can create a recurrent link from the output to a merger where it joins with the signal from a data source.

I’d recommend the former though as gamma memories are more flexible in terms of memory depth and resolution than plain feedback loops.

Including the temporal dynamics can indeed be useful - the only reason why it wasn’t included in the tutorial is because I wanted to keep things simple.

Pete

]]> By: IvanR http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/comment-page-1/#comment-15 IvanR Sat, 03 Jun 2006 03:57:14 +0000 http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/#comment-15 Could this be used in some way with time-dynamic nets? I reckon that considering how the trading unfolds in time could be useful. Could this be used in some way with time-dynamic nets?

I reckon that considering how the trading unfolds in time could be useful.

]]> By: JoLex http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/comment-page-1/#comment-14 JoLex Thu, 01 Jun 2006 23:40:00 +0000 http://blog.peltarion.com/2006/06/01/hebbian-novelty-filters-for-financial-analysis/#comment-14 Thx. Just what the doctor ordered. Thx. Just what the doctor ordered.

]]>