netrome / DeepGeneration

Generation of synthetic data sets using generative adversarial neural networks
0 stars 0 forks source link

Evaluation measures #33

Open netrome opened 6 years ago

netrome commented 6 years ago

The method should not be able to improve JD error or euclidean distance. It is common in the litterature to report the percentage of the recordings that falls below a specific range. Perhaps I should consider something like this.