... Hello Guy I've been using ANN's for a while, having studied them over several years with the OU here in the UK. If you need any specific help, let me know...
Hello, I have created this Yahoo list so that users of EasyNN-plus can exchange ideas. However, the list is proving to be unreliable because of one or two ...
I have sent invitations to join this group to all registered users of EasyNN-plus, EasyNN and Neural Planner. Should I make the group available to anyone by...
... Answering half my own question: It seems groups are listed in the Yahoo directory regardless of how they are created. Steve -- EasyNN-plus. The easy way to...
Hello All, The input data that I use is very noisy. Would you think it best to check control and decay within the learning controls using EasyNN? John H....
Good idea to set up a Yahoo group Steve Lots of experiences now with easy NN, Problem is the race horses just dont seem to recognise AI, when they start...
... Do you mean learning rate decay? Does the network oscillate? Learning rate decay will normally stop a network oscillating, even if the data is very noisy....
... OK, the door is open. Anyone can join by clicking the button on the messages area of the web site or by finding us in the Yahoo directory. Steve -- ...
... Validating examples can be used to stop learning before the network is overtrained. Validating can stop learning when the network has learned enough to...
Steve wrote, ... I do. ... Not to my knowledge. What do oscillations look like exactly? Probably best to train a network unchecked to see what happens. John...
... John, Look at the graph. The average error will be unsteady and usually the max error will have spikes. Steve -- EasyNN-plus. The easy way to build neural...
Hello All, How does one cope with zero's contained within one's data? E.g. if a horse racing trainer had 0 winners, and another trainer 5 winners, will the ...
... John, In your first question I assume the number of winners a trainer has had does have some influence on the future performance. More winners would...
Afternoon all, Nice to see some other members of the group have an interest in horse racing I prefer to use lengths beaten rather than finishing position. A...
Paul Simmons
prs@...
Jul 28, 2003 4:03 pm
25
Paul wrote, ... I think there is a lot of potential. It's a matter of getting the right variables, and the mix right? ... I have to agree somewhat. ... Are...
Does anyone know if there is a current upgradefor Easynn Plus version 1.00? In other words, what is the latest paid version for Easynn Plus? 29 July 2003 Guy...
Hello All, Although this is not strictly related to EasyNN (hope Steve doesn't mind non-related queries). What is the best method to present one's data. E.g. ...
good questions I am also pondering John, It seems to me that the number of input columns must have some relationship to the minimum number of traing samples...
... It depends on what you are trying to do. It you are trying to get a low error then you need five to ten times as many examples as inputs but this only...
Steve, My inputs are the five components; open, high, low, close, volume, each with 25 samples in the same "bell" shaped time interval; i.e. samples are before...