An Efficient Frequency Encoding Scheme for Optical Convolution Accelerator
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsDear Editor,
The authors have presented an encoding scheme for utilizing the computational resources of optical convolution accelerators. The manuscript has been well-designed and the topic is interesting. Moreover, the results support the title and abstract. So, I recommend a minor revision based on the following comments:
1. Discuss the numerical results in the abstract. Simply writing the quantitative results is not enough.
2. No noise was observed in this work. What is the signal-to-noise ratio?
3. What is the data set of this work? How many data points were used to evaluate the proposed method?
4. In order to better evaluate the results obtained, a comparison table with other methods needs to be added to the manuscript.
Kind regards
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper introduces a new frequency encoding scheme for optical convolution accelerators, offering a substantial increase in computational efficiency for two-dimensional matrix convolutions. The proposed approach is innovative and well-supported by theoretical analysis and experimental results. It has sound implications for the field of optical computing, particularly in high-performance applications like neural network accelerations and deep AI training.
While the work is of high quality and deserving of publication, a few areas require minor improvements to enhance clarity and completeness.
1. The quality of some figures could be imrpoved for better clearity. Like figure 1 and 3.
2. The article references the encoding scheme by Xu et al. Include a more detailed quantitative and qualitative comparison, discussing scenarios where the proposed method might not outperform existing approaches (e.g. limitations in hardware scalability).
3. It would be valuable for the authors to elaborate on the error analysis. For example, what impact might noise or system imperfections have on scalability and accuracy? This will provide readers with a fuller picture of the method’s practical applicability.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe manuscript ID photonics-3360945 mainly presents a study about particular optical convolution accelerator for the assistance of a frequency encoding system. Please see below a list of comments to the authors:
1. From the presentation of the methodology is not clear how were selected the parameters proposed for the system studied.
2. The temporal response of the system proposed should be estimated.
3. Is there a potential influence of incident polarization in the performance of the system?
4. A photo of the experiment would be welcome.
5. The manufacturer and model of the subsystems employed must be provided.
6. Advantages and disadvantages of this frequency encoding scheme for optical convolution accelerator should be summarized in the discussion section taking into account different alternatives. The authors are invited to see for instance: and https://doi.org/10.1364/OE.456255
7. The main results should be confronted with updated publications in the topic. You can see for instance: https://doi.org/10.1016/j.rinp.2024.107968
8. The perspectives described could be improved.
9. Experimental error bar must be reported.
10. It is suggested to split the citations presented in collective form in order to better justify each reference selected for this report.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsI appreciate the effort of the authors to clarify the points raised in the initial review stage; however, most of fundamental issues are still present in the reviewed version of the manuscript, please see below:
*The methodology to describe how were selected the parameters proposed for the system studied is unclear.
* The temporal response of the system proposed was not estimated.
*The potential influence of incident polarization in the performance of the system was not better described.
*The photo of the experiment was not included.
*In the analysis of the advantages and disadvantages of the frequency encoding scheme has ommited confrontation with alternatives as suggested.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 3
Reviewer 3 Report
Comments and Suggestions for AuthorsI agree with the reviewed version of the manuscript. In my opinion, this work can be considered for publication in present form.