Multiple Back-Propagation (with CUDA) / News: Recent posts

Version 2.2.1

Multiple Back-Propagation v. 2.2.1 is now available

This version corrects a bug that prevented data files in a format other than CSV to be properly read (As a result MBP would crash when training a network). When using CUDA the value of the max step size (configuration) is now taken into account (in previous versions the value of 10.0 was used regardless of the max step size specified).

Multiple Back-Propagation Version 2.2... read more

Posted by Noel Lopes 2010-05-24

Version 2.1.4 is now available

Version 2.1.4 corrects the following problems (bugs):
- When using CUDA to train a neural network, if more than one layer
contained selective activation neurons, the train would fail (fixed).
- After successfully training a network (using CUDA) it was no longer
possible to train new networks (fixed).

Posted by Noel Lopes 2010-03-03

Multiple Back-Propagation 2.1.3 available

Version 2.1.3 corrects the following problems (bugs):

  • When changing tabs during the training procedure, sometimes the program would crash (fixed).
  • The robust training under CUDA (which was the default for people with NVIDIA devices) had a bug which resulted in the network not being properly trained (fixed).
  • When training under CUDA the current epoch is now displayed (the total time of training is still not displayed correctly).
Posted by Noel Lopes 2010-02-01

Multiple Back-Propagation Version 2.1.2a

Multiple Back-Propagation version 2.1.2 had a bug that prevented it from working under Windows XP and earlier systems. Version 2.1.2a fixes this problem.

Posted by Noel Lopes 2010-01-10

MBP 2.1.2 is now available

Version 2.1.2 is now available for download. This version supports CSV (Comma-separated values) files. CSV files can easily be generated by many programs such as Excel, thus CSV files are now the preferred method of input data (the other formats remain only for compatibility purposes) to Multiple Backpropagation. Furthermore, reading from CSV files was optimized and should be faster than reading data from other formats (especially for large datasets). Moreover when reading this format Multiple Back Propagation supports your locale decimal and list separators (there is no longer need for changing your data).... read more

Posted by Noel Lopes 2010-01-08

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

JavaScript is required for this form.

No, thanks