Incorporating Derivative-Free Convexity with Trigonometric Simplex Designs for Learning-Rate Estimation of Stochastic Gradient-Descent Method

Round 1

Reviewer 1 Report

The authors propose a new definition of convexity for computing learning rate schedulers for SGD methods in this paper.

It is shown that the new learning rate scheduler can be used for other convex optimization applications involving deep learning and pattern analysis.

These research results merit publication as a paper.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

I just have one question. What is the basis of Eqs. (15) - (18)? Is it special or representative?It is highly recommended to add supplementary statement, otherwise it will be difficult for the reader to believe that the algorithmic advantages shown in Figures 1-4 are objective rather than specific.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

I reviewed the manuscript tittle Incorporating Derivative-free Convexity with Trigonometric Simplex Designs for Learning Rate Estimation of Stochastic

Gradient Descent Method

 

the paper is well written and organized properly. It is easy to follow and understand in most parts. The technical sound of the paper is substantial However, authors need to consider the following comments to improve the quality and sounds of the paper

 

Introduction

Authors need to provide some more explanation and general information in the first part of the introduction before starting with the equations to facilitate better understanding for readers

Provide some support/references for the first paragraph and where the information has been sourced,

Authors need to check the citation and make sure all references have been cited correctly

 

Method/approach

It is not clear in the paper as these came all under the introduction, the paper should have a clear and separate part for the methodology. And why such methodology/ approach was used. This is a very important part for other to utilize this work

 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf