search

FlorentF9 / DeepTemporalClustering

:chart_with_upwards_trend: Keras implementation of the Deep Temporal Clustering (DTC) model
MIT License
219 stars 58 forks source link

Training and Validation Losses #16

Open nikosphys opened 3 years ago

nikosphys commented 3 years ago

Can you please explain the losses: L, Lc, Lr and T

FlorentF9 commented 3 years ago

Lc is the clustering loss (soft hardening loss using KL divergence on the soft assignments), Lr is the autoencoder reconstruction loss (MSE) and L = Lr + gamma * Lc is the total loss, where gamma is a weighting hyperparameter. T stands for nothing and should be removed.