XTRACTORAI

Goal oriented loss functions to achieve targets as opposed to ad-hoc tuning of metrics... Neural architectures with complementary loss functions to avoid redundant learning... Justify model predictions using your domain knowledge supplemented by our interpretable neural features