Semi-Supervised Deep Time-Delay Embedded Clustering for Stress Speech Analysis

Round 1

Reviewer 1 Report

The pape presents a semi-supervised clustering method (SDTEC) based on their previous unsupervised one, i.e., DTEC, by incorporating the prior knowledge of pairwise constraints in embedding space and simultaneously perform the clustering assignments.

I feel that the paper is quite similar to peng's work, Structured AutoEncoders for Subspace Clustering, TIP18; deep subspace clustering with sparsity prior, IJCAI16. However, the relations and difference did not analyzed.

The ablation study should be conducted to verify the role of each term of your loss.

The baselines seems quite old. It would be better to compare with SOTA apporaches.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript has to do with quite topical research. It introduces an exciting idea connected with a sufficient level of mathematical background.

Although there are some remarks.

An extensive correction of English grammar and style is required.

Abbreviations should be put somewhere at the beginning of the paper.

The paper’s structure is missing in the Introduction.

The paragraph, which starts with “In machine learning applications…” is written in a non-understandable way (quite messy). Please rewrite that properly.

The experimental part should be extended. Some more comparison of analogs and their analysis should be added.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I have no more suggestions.