Closed accosmin closed 8 years ago
Batch trainer fails before overfitting. To investigate if the line-search methods are robust enough to handle the "non-convexity" present when training MLPs/convnets.
Batch trainer fails before overfitting. To investigate if the line-search methods are robust enough to handle the "non-convexity" present when training MLPs/convnets.