how to focus on positive distances?

Speaker-Identification / You-Only-Speak-Once

Deep Learning - one shot learning for speaker recognition using Filter Banks

142 stars 36 forks source link

how to focus on positive distances? #3

Open VictorBeraldo opened 3 years ago

VictorBeraldo commented 3 years ago

Thanks for the repo!

I have a question... you've said that "Hence our model is optimized to maximize negative accuracy". Can you help me how to maximize positive accuracy in this your code?

vivekkr12 commented 3 years ago

Take a look here - you can adjust the threshold.