Service notice, 15 May 2026: Our services are currently undergoing maintenance. Previously submitted results remain available, but we are unable to accept new job submissions at the moment. We apologise for the inconvenience and are working to restore the services as soon as possible.

Confidence

The highest scoring state from the outputs from the neural network was taken, and the score of the next highest predicted state subtracted:

$$Confidence = integer (10 \times (out_{max}-out_{next}))$$ (1)

This is similar to the method exploited by PHD [17].