ICACT20260055 Question.1
Questioner: abilaaniloshan@gmail.com 2026-02-09 오후 3:45:14 |
ICACT20260055 Answer.1
Answer by Auhor b.emmanuelkwesi@lancaster.edu.gh 2026-02-09 오후 3:45:14
Chrome  Click!! |
| Why did you choose a hybrid CNN–Vision Transformer architecture instead of using only CNNs or only Transformers for PDF malware detection? |
|
ICACT20260055 Question.10
Questioner: vicky940318@gmail.com 2026-02-11 오후 8:05:46 |
ICACT20260055 Answer.10
Answer by Auhor b.emmanuelkwesi@lancaster.edu.gh 2026-02-11 오후 8:05:46
Chrome Click!! |
| What are the main limitations of the proposed system, and how could future work address these limitations to improve real-world deployment? |
|
ICACT20260055 Question.2
Questioner: arjunvirvashisht@gmail.com 2026-02-10 오전 12:50:29 |
ICACT20260055 Answer.2
Answer by Auhor b.emmanuelkwesi@lancaster.edu.gh 2026-02-10 오전 12:50:29
Chrome Click!! |
| Do you plan to test this model against obfuscated or encrypted PDF malware in the future? |
|
ICACT20260055 Question.3
Questioner: arjunvirvashisht@gmail.com 2026-02-10 오전 12:51:24 |
ICACT20260055 Answer.3
Answer by Auhor b.emmanuelkwesi@lancaster.edu.gh 2026-02-10 오전 12:51:24
Chrome Click!! |
| Was the training dataset for this study obtained from a public repository? |
|
ICACT20260055 Question.4
Questioner: hominpark@etri.re.kr 2026-02-10 오전 11:01:01 |
ICACT20260055 Answer.4
Answer by Auhor b.emmanuelkwesi@lancaster.edu.gh 2026-02-10 오전 11:01:01
Chrome Click!! |
| Could the 99% accuracy achieved using only the PDFRep dataset without augmentation indicate overfitting to specific dataset biases? Also, regarding the avg_degree outlier identified in the SHAP analysis, did you investigate if this represents a potential evasion technique? |
|
ICACT20260055 Question.5
Questioner: mrajhans@paloaltonetworks.com 2026-02-11 오전 7:24:18 |
ICACT20260055 Answer.5
Answer by Auhor b.emmanuelkwesi@lancaster.edu.gh 2026-02-11 오전 7:24:18
Chrome Click!! |
| How robust is the proposed CNN + ViT model to adversarially crafted or heavily obfuscated PDF malware (e.g., structure-preserving attacks or feature-manipulation evasion), given that evaluation is limited to the PDFRep dataset without explicit adversarial testing? |
|
ICACT20260055 Question.6
Questioner: vicky940318@gmail.com 2026-02-11 오후 8:05:03 |
ICACT20260055 Answer.6
Answer by Auhor b.emmanuelkwesi@lancaster.edu.gh 2026-02-11 오후 8:05:03
Chrome Click!! |
| Why did the authors choose to combine Convolutional Neural Networks (CNN) with Vision Transformers (ViT) instead of using a single deep learning model for PDF malware detection? |
|
ICACT20260055 Question.7
Questioner: vicky940318@gmail.com 2026-02-11 오후 8:05:13 |
ICACT20260055 Answer.7
Answer by Auhor b.emmanuelkwesi@lancaster.edu.gh 2026-02-11 오후 8:05:13
Chrome Click!! |
| How does the proposed hybrid CNN–ViT model improve robustness against obfuscation techniques and zero-day PDF malware compared to traditional machine learning approaches? |
|
ICACT20260055 Question.8
Questioner: vicky940318@gmail.com 2026-02-11 오후 8:05:24 |
ICACT20260055 Answer.8
Answer by Auhor b.emmanuelkwesi@lancaster.edu.gh 2026-02-11 오후 8:05:24
Chrome Click!! |
| What role does the PDFRep dataset play in achieving high detection performance, and how does dataset size and diversity influence the model’s effectiveness? |
|
ICACT20260055 Question.9
Questioner: vicky940318@gmail.com 2026-02-11 오후 8:05:35 |
ICACT20260055 Answer.9
Answer by Auhor b.emmanuelkwesi@lancaster.edu.gh 2026-02-11 오후 8:05:35
Chrome Click!! |
| How does the SHAP explainability framework help identify the most influential features in the model’s decision-making process? |
|
