Categories : Neural Network Programs
Backprop is a standalone multi-layer neural network simulator that is based upon the popular backpropagation learning algorithm. The goal of this simulator is to provide users with a friendly and easy to use environment for experimenting with backpropagation networks. To achieve this, I put a lot of effort into making the user interface give as much visual feedback as possible, especially during network training, as well as giving the user easy to use interfaces for changing the attributes of the network, such as learning rates, momentum, and so forth. You can zoom in on the network graphically to see weight values in more detail, or zoom out in order to make visible larger, more complicated network architectures. You can speed up, or slow down, the rate at which error graphics and network state are updated during training. It is features like this that I hope will make Backprop your first choice for experimenting with backpropagation neural networks.
Backprop displays activation and weights during training as they change, and allows the user to enable/disable/configure the use of momentum and learning rate during training. You can also enable or disable a bias term to see what effect it has on convergence during training. Finally, it allows you to specify the use of sigmoid or htan activation functions.
OS : Microsoft Windows 95/98/ME/NT/2000/XP
Owner : CTS
Prize :
File Size : 76.32 KB
Lisensi : Shareware
Website / Download
http://www.users.cts.com/crash/s/slogan/backprop.html
Tidak ada komentar:
Posting Komentar