Need to add code for throwing an exception when the weight values or input/output values exceed the range

Aplokodika / MachineLearning

The main motive is to create a good tool for conducting AI experiments in vast areas.

2 stars 5 forks source link

Need to add code for throwing an exception when the weight values or input/output values exceed the range #16

Closed sreramk closed 8 years ago

sreramk commented 8 years ago

This is to ensure that the weight values, and the input/output values don't get stuck at Nan (not a number) or infinity. If such an exception isn't handled, the system may seem to run fine without any problems initially, but soon, the influence of this error would start becoming significant. So such an issue exception must be handled within the update function.

sreramk commented 8 years ago

This issue greatly depends on issue #12. It is because, the exceptions should no longer be thrown because of exceeding float value's bounds.

sreramk commented 8 years ago

This issue was solved in the the latest upload: https://github.com/Aplokodika/MachineLearning/commit/137c11dd957472d8679fce1024fc713efb8606c3#diff-89ff528a30f91fbfd47b4ec3be3c38bf