I was just checking a script and notice that I have forgotten to change the "if" commands to "itf" in the sample files. Both commands work and do the same...
A problem has been found with the nested script command parser. If a script has a command with spurious spaces they will be removed. Unfortunately this can...
Hi Steve, After working with your EasyNN for a while this is my wish list for Ver6:: 1) Import a xxx.csv file but set default to last Numeric columnas TARGET...
The wish list that Marc posted has some interesting ideas and they have been added then to the list for possible inclusion in future versions of EasyNN-plus. ...
... So far as I am aware all the formulas for the number of hidden nodes and hidden layers fail in some way. The only method that I know works is the one that...
Mistakes: I really should have said it was a bug in 5.0d but for some reason I was stuck in the world of version 4.0 ... It is fixed in 5.0e which is now...
I am have trouble with my e-mail messages being rejected again. For example hanmail.net are rejecting all of them. This includes registration messages. If you...
The wish list for version 6 is a bit lacking! So far the only ideas are user defaults for importing and a macro recorder/player to produce scripts. Both these...
Hi Steve, I am not quite sure if easynn is using the backpropagation algorithm but, if it is using it then can that supercede with some new faster algorithms...
Well it would be nice if you had: 1) A method to do 10xCV automatically with 10 random splits of the data with the ability to output the predicted values...
A results grid with a Save As feature, that saves the selected result columns to a delimited file, with maybe an auxiliary file that contains auxiliary ...
... Darshan, RPROP has some promise. I read quite a lot about it when it was first developed in the early 90's but have never tried to implement it. My first...
... Have you tried Action > Leave Some Out ? It could be changed to do random splits. Bug: at present it only works once without reloading the file. Subsequent...
Re: 1) the classics for cross validation are 10xCV and moving window which you have already BUT the 10xCV would be much faster to complete. It really fills in...
... Did you miss the save as "Do you want to use comma separators?" prompt in File Query? It defaults to "No" so you may always be producing text results...
... separators?" prompt in File Query? It defaults to "No" so you may always be producing text results files. *.csv files are linked to Excel and will open...
I have started working on the macro facility for version 6. So far the macro is just a record of calls to major functions as they occur. For example below is...
I like the unified macro/script approach... You may consider a small "translation" utility to convert scripts to the macro syntax and then only one scripting...
I guess I am now sure what the difference is between the macro facility and script and why you would need both. If you can just capture your keystrokes so it...
... Unification will probably be the best approach. It will be easier to call the low level script commands from macros rather than convert scripts to macro...
The default installation of some recent versions of EasyNN-plus is in the folder dc:\Program Files\EasyNN-plus rather than dc:\Program Files\EasyNN The...
Some progress has been made with macros. The following macro commands were produce by recording. This macro works OK when played. Nothing was typed. Lines...
This looks really usable to me. I was wondering about the repeated line "ChangeNetwork(1, 1) 16293849" I assume these extra lines can be deleted in the final...
... ChangeNetwork(1, 1) is a low level function that says increment hidden layer 1 by 1. It is called during optimization until the optimum number of nodes is...
Steve, I'm probably not going to have the time to come up with suggestions, this time around, though there will probably be time to beta-test version 6 when...
... Dwayne, It will be great if you can help with beta testing. Your scripts were very helpful last time. Steve -- EasyNN-plus. The easy way to build neural...
Macro progress. The information below is copied from the Help file so you can see what is going to be available. I think those of you who use scripts will...
The first beta release of EasyNN-plus 6.0a is ready for testing. The additions and changes are as follows:- 1. New Macro record and playback facility...
So far no problems have been found with beta 1. This means that either it is free of problems or more testers are needed. I would like to think beta 1 is...