Leverage Boosting and Transformer on Text-Image Matching for Cheap Fakes Detectionâ€
, Minh-Son Dao 3,*, Duy-Dong Le 4, Kim-Phung Thai 4, Quoc-Hung Nguyen 4 and Thuy-Kieu Phan-Thi 4Round 1
Reviewer 1 Report
1. The abstract part could be improved. A good abstract should be attractive.
2. Please carefully do proofreading before resubmission. There are several mistakes, like typos, in this article. For example, in line 119, “Datase” must be corrected.
3. The experimental part is not adequate. It should include more results and discussion, particularly, in the failure cases.
4. The description of the proposed boosting algorithm, like lines 245–252, could be improved. It is unclear to understand the details.
5. Please reproduce Fig. 7. The text in this figure is too small.
6. The conclusion part failed to highlight the significance of this study. It could be improved.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This manuscript proposed several methods combine text-image matching and image captioning methods via ANN/Transformer boosting schema to classify a triple of (image, caption1, captions) into OOC (out-of-context) and NOOC (no out-of-context) labels to detect cheapfakes. As such, the matter is of interest. However, the manuscript has several limits:
1. It seems that more references need to be read.
2. Line 11, It is suggested to add an explanation in the manuscript: the difference between "caption1" and "caption".
3. Line 36:” The blossoming of deep learning has opened new domains and technology, one of which is Deepfake.” It is suggested to add references to increase persuasiveness.
4. Line116: "By combining several technologies...", it is recommended to list these technologies specifically.
5. Line 168, is equation (1) proposed by yourself? If it is quoted, please add references. If it is proposed by yourself, it is recommended to introduce the "margin" item detailed.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I think all the commefnts have been addressed.
Reviewer 2 Report
The paper merits publication.
