Dear Reviewers, we thank you for taking the time to provide valuable feedback. Below we address the main issues raised. Its performance depends on our ability to predict the distribution over future frames with low entropy. We will emphasize these aspects more in a revised version. RNNs to model dynamics in the latent space.

We thank all the reviewers for their insightful and encouraging comments. Per your suggestion, we will update the appendix by adding more explanations about the proof ideas. Similarly, we can extend convergence results in Theorem 4 in Appendix from FS to IFS. We expect that the approach developed in this paper will fuel this future investigation. We will update this into the revision.

We thank all the reviewers for their insightful and encouraging comments. Per your suggestion, we will update the appendix by adding more explanations about the proof ideas. Similarly, we can extend convergence results in Theorem 4 in Appendix from FS to IFS. We expect that the approach developed in this paper will fuel this future investigation. We will update this into the revision.

Experiment Details: Figure 3 shows the general architecture of FQF. 'inefficient hyperparameter' we mean the atom locations (uniformly distributed between -10 and 10) in C51, quantiles FQF requires only the number of quantiles. We leave the theoretical analysis and comparison between IQN and FQF to future work. We will check if we can figure out why and explain it in detail. As shown in our toy case, FQF does achieve better distribution approximation.