ICACT20230167 Question.1
Questioner: lp_saikia@yahoo.co.in 2023-02-19 오후 5:41:47 |
ICACT20230167 Answer.1
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-19 오후 5:41:47
Chrome  Click!! |
| Why the number of features is 78 as shown in the left picture(slide no.7)? |
Thank you for your question, as these 78 features are the default output of cicflowmeter. |
ICACT20230167 Question.10
Questioner: sdk@etri.re.kr 2023-02-20 오후 7:56:59 |
ICACT20230167 Answer.10
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-20 오후 7:56:59
Chrome Click!! |
| First question: Please tell us more about the system specifications of the AI BOX used in the experiment.
Second question: It seems that agent software must be installed on the server (or PC) to provide input data for the system combined with AI BOX and embedded IDS. Have you considered this? |
Thank you for your questions. The AI box uses a CPU developed by MediaTek, model i500, with 4GB of ram, 32GB of hard drive, and the Yocto Linux operating system. If we learn more about the AI box, we can share it with you later. We did not need additional agent software to receive the input data during the experiment because we were directly listening to the ai box network card's packet traffic.
|
ICACT20230167 Question.11
Questioner: 6025211020@mhs.its.ac.id 2023-02-21 오후 12:26:51 |
ICACT20230167 Answer.11
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-21 오후 12:26:51
Chrome Click!! |
| Interesting study, feature selection phase caught my attention. Can you explain why ANOVA and SelectKBest were chosen to perform feature selection over many other feature selection techniques? Is it because this technique is the most effective and efficient? Thank You. |
Thank you for your question. Because we have experimented and found that this method reduces the number of input features and maintains good results. |
ICACT20230167 Question.2
Questioner: lp_saikia@yahoo.co.in 2023-02-19 오후 5:45:13 |
ICACT20230167 Answer.2
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-19 오후 5:45:13
Chrome Click!! |
| To analyze the hardware function information, what tool you are using? |
Thank you for your question, the hardware information is provided by ITRI, we do not use any tools |
ICACT20230167 Question.3
Questioner: lp_saikia@yahoo.co.in 2023-02-19 오후 5:13:07 |
ICACT20230167 Answer.3
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-19 오후 5:13:07
Chrome Click!! |
| What are the limitations of your work? |
Thank you for your question, the current limitation of this study is mainly the hardware support part, when the hardware can not use docker and other packages, may cause the system can not be used. |
ICACT20230167 Question.4
Questioner: nvlinh@cs.ccu.edu.tw 2023-02-19 오후 5:15:21 |
ICACT20230167 Answer.4
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-19 오후 5:15:21
Chrome Click!! |
| What are your contributions? |
We have built an automated intrusion detection system that is highly accurate and can be easily extended to other embedded devices. |
ICACT20230167 Question.5
Questioner: nvlinh@cs.ccu.edu.tw 2023-02-19 오후 5:16:18 |
ICACT20230167 Answer.5
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-19 오후 5:16:18
Chrome Click!! |
| What is inside AIbox, e.g., deep learning models, data pre-processing? |
Thank you for your question, the whole intrusion detection system. |
ICACT20230167 Question.6
Questioner: nvlinh@cs.ccu.edu.tw 2023-02-19 오후 5:17:34 |
ICACT20230167 Answer.6
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-19 오후 5:17:34
Chrome Click!! |
| How about the response time per request if the incoming traffic is too heavy. |
We have not yet conducted experiments on this part, and if we find anything later, we can share it with you. |
ICACT20230167 Question.7
Questioner: m11007503@gapps.ntust.edu.tw 2023-02-19 오후 5:19:38 |
ICACT20230167 Answer.7
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-19 오후 5:19:38
Chrome Click!! |
| Did you encounter any difficulties during the research? |
Thank you for your question, the training of the models and the pre-processing of the data, as well as the deployment of the intrusion detection system to AI BOX, were all difficult points, but we overcame them successfully. |
ICACT20230167 Question.8
Questioner: m11007503@gapps.ntust.edu.tw 2023-02-19 오후 5:40:11 |
ICACT20230167 Answer.8
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-19 오후 5:40:11
Chrome Click!! |
| Is there anything that can be done to improve this study in the future? |
Thank you for your question, at present, we can try to use deep learning technology to reduce the error rate is one of the directions we can work on in the future. |
ICACT20230167 Question.9
Questioner: m11007505@mail.ntust.edu.tw 2023-02-20 오후 5:46:29 |
ICACT20230167 Answer.9
Answer by Auhor lchen@mail.ntust.edu.tw 2023-02-20 오후 5:46:29
Chrome Click!! |
| Did you run the system in a real environment and how efficient is it? |
Thank you for your question, we have not yet tested this system in a real-world network traffic environment, but we will share with you if we conduct further research in the future. |
