6. AutoSave

Fig. 1. Use the AutoSave function to set criteria for automatic saving of networks.
Use this function to save the networks automatically during training. All the control parameters and the interconnection weights will be saved. The file extension will be incremented as the network is saved. If the network file name is "XXX.GA", the first auto-saved network will be "XXX.001", and the next will be "XXX.002" and so on. You are allowed to change the file name, extension and directory path.
You can select any combination of the criteria, and the network will be saved when any one of them is met:
- AutoSave Based on Error
- Automatically save the network when the error drops by XXX % from the last saving. You can make this amount self-adjustable by clicking on the radio button next to it.
- AutoSave on Every YYY Iterations
- Automatically save the network on every YYY iterations. The option is useful if you want to leave the training unattended for a long period of time.
Main Page
1. A Quick Tour of NeuroForecaster and GENETICA
2. Neural Network Applications
3. Genetic Algorithms & Genetically Evolved NNs
4. Using GA To Reduce Input Data Dimension
5. AutoTest Function
6. AutoSave Function
7. AutoStop Function
8. VisuaData - A Neural Net Preprocessor for Traders
9. References
10. Free financial data & info, links to financial pages
Papers Available:
1. NeuroGenetic Computing
2. NeuroFuzzy Computing
3. Select! (Technical Report I): Trading With A Stock's Alpha
4. Select! (Technical Report 2): Trading for the Risk-Averse
Copyright © 1993-96, NIBS Pte Ltd.