What is the meaning of labels?

[zhouhao-learning]

I see your data_generator function, but labels = np. empty ((input_data [0]. shape [0], 2 * output_size). Astype ('float32'), which is not real labels, is there any point in defining labels like this? So what is my model learning? Can you give me some specific explanations? Thank you very much.

[pasqualil]

The labels are required by Keras loss function format. We use empty labels as VAMPnets are a case of unsupervised learning since the loss function can be evaluated using only the training data.

[zhouhao-learning]

@Obudo Well, thank you very much! Another question. What does lag mean in the paper?

[amardt]

Hey, lag is another word for the $\tau$ timestep from the paper. It means how many timesteps you want to predict in the future, broadly speaking. You will always train a model to approximate the time propagator K for a specific timestep. Therefore, you split your data into $traj(t)$ and $traj(t+\tau)$. The first trajectory is then $\tau$ lagging behind, so to say.

Hope this makes it clearer! Best Andreas

[zhouhao-learning]

@amardt OK, thank you for your guidance.