Next Article in Journal
A Simplified Algorithm for Dealing with Inconsistencies Using the Analytic Hierarchy Process
Next Article in Special Issue
Asynchronous Gathering in a Dangerous Ring
Previous Article in Journal
Vineyard Gap Detection by Convolutional Neural Networks Fed by Multi-Spectral Images
Previous Article in Special Issue
Recent Developments in Low-Power AI Accelerators: A Survey
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Algorithms
Volume 15
Issue 12
10.3390/a15120441
Review Report
algorithms-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
Article
Peer-Review Record

A Methodology to Design Quantized Deep Neural Networks for Automatic Modulation Recognition

Algorithms 2022, 15(12), 441; https://doi.org/10.3390/a15120441
by David Góez 1,2, Paola Soto 1,2, Steven Latré 1, Natalia Gaviria 2 and Miguel Camelo 1,*
Reviewer 1:
Wenwu Xie
Reviewer 2:
Guan Gui
Algorithms 2022, 15(12), 441; https://doi.org/10.3390/a15120441
Submission received: 19 October 2022 / Revised: 14 November 2022 / Accepted: 18 November 2022 / Published: 22 November 2022
(This article belongs to the Collection Parallel and Distributed Computing: Algorithms and Applications)

Round 1

Reviewer 1 Report

Q1. The authors proposed a quantitative deep neural network design method, which was verified on VGG10 1D-CNN model, and good results were obtained. Further testing of other similar models is recommended to make the results more convincing.

Q2. In order to further verify the superiority of this quantitative design method. Has the author considered comparing the quantization algorithm designed by using the quantitative design method in this paper with the existing quantization algorithm?

Q3. For the aquamarine color points in Figure 4, in addition to the two points (i2,a2,w2 and (i8,a8,w8), other representative points should also be marked to show the regularity.

Q4. Figures 8 and 9 show the results of the different quantized versions of the model. Is this result actually deployed in an FPGA experiment?

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper proposed a quantized deep neural networks-based automatic modulation recognition method. Both the technical contribution and the presentation are acceptable.  The reviewer has a main concern about the novelty.  There are much deep learning-based AMR or AMC methods that have been proposed in recent years, such as: 

Multitask-Learning-Based Deep Neural Network for Automatic Modulation Classification. IEEE Internet Things J. 9(3): 2192-2206 (2022)

Automatic Modulation Classification Using Gated Recurrent Residual Network. IEEE Internet Things J. 7(8): 7795-7807 (2020)

Lightweight Automatic Modulation Classification Based on Decentralized Learning. IEEE Trans. Cogn. Commun. Netw. 8(1): 57-70 (2022)

NAS-AMR: Neural Architecture Search-Based Automatic Modulation Recognition for Integrated Sensing and Communication Systems. IEEE Trans. Cogn. Commun. Netw. 8(3): 1374-1386 (2022)

What is the main difference between these works? Hence, it is better to highlight the novelty of the proposed method in this paper.

The algorithm statement of the proposed method should be given. Also, there are some format problems and typos. Please check the whole paper.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The manuscript is well revised according to the comments of reviewers, and it can be accepted in its current version.
Back to TopTop