ICACT20260019 Question.1
Questioner: abilaaniloshan@gmail.com 2026-02-09 오후 12:56:31 |
ICACT20260019 Answer.1
Answer by Auhor a.rahman@mst.edu 2026-02-09 오후 12:56:31
Chrome  Click!! |
| Why did you choose classical machine learning models instead of deep learning approaches for SmartNIC-based intrusion detection? |
|
ICACT20260019 Question.2
Questioner: abilaaniloshan@gmail.com 2026-02-09 오후 12:57:11 |
ICACT20260019 Answer.2
Answer by Auhor a.rahman@mst.edu 2026-02-09 오후 12:57:11
Chrome Click!! |
| Since NSL-KDD is an older dataset, how applicable are your results to modern high-speed networks? |
|
ICACT20260019 Question.3
Questioner: qga0809@chungbuk.ac.kr 2026-02-09 오후 1:21:16 |
ICACT20260019 Answer.3
Answer by Auhor a.rahman@mst.edu 2026-02-09 오후 1:21:16
Chrome Click!! |
| Your results show that Random Forest achieves 100% accuracy with 0.0114ms per-sample latency, while Decision Tree has only 0.0002ms latency. For SmartNIC deployment handling 100Gbps traffic, what is the maximum packets per second that Random Forest can process, and would this meet real-world data center requirements? |
|
ICACT20260019 Question.4
Questioner: ygt991129@mail.ustc.edu.cn 2026-02-09 오후 2:01:28 |
ICACT20260019 Answer.4
Answer by Auhor a.rahman@mst.edu 2026-02-09 오후 2:01:28
Chrome Click!! |
| How can classical machine learning classifiers be adapted to real-time changing network traffic environments to enhance the robustness of intrusion detection? |
|
ICACT20260019 Question.5
Questioner: ygt991129@mail.ustc.edu.cn 2026-02-09 오후 2:01:45 |
ICACT20260019 Answer.5
Answer by Auhor a.rahman@mst.edu 2026-02-09 오후 2:01:45
Chrome Click!! |
| Under the strict resource constraints of SmartNICs, how can algorithms like Random Forest be further optimized to reduce latency without compromising accuracy? |
|
ICACT20260019 Question.6
Questioner: Lee19711936516@163.com 2026-02-09 오후 2:33:52 |
ICACT20260019 Answer.6
Answer by Auhor a.rahman@mst.edu 2026-02-09 오후 2:33:52
Chrome Click!! |
| In the real-time intrusion detection scenario based on SmartNIC, Random Forest (RF) exhibits the best balance between accuracy and inference time compared to Decision Tree (DT) and K-Nearest Neighbor (KNN). What are the key points in its core algorithm design and hyperparameter optimization strategy that make it adapt to the resource constraint characteristics of SmartNIC? |
|
ICACT20260019 Question.7
Questioner: Lee19711936516@163.com 2026-02-09 오후 2:34:56 |
ICACT20260019 Answer.7
Answer by Auhor a.rahman@mst.edu 2026-02-09 오후 2:34:56
Chrome Click!! |
| This study emphasizes the importance of data preprocessing for optimizing model performance on the SmartNIC platform. For the NSL-KDD dataset, operations such as correlation feature filtering (removing feature pairs with r>0.8) and category feature hot encoding were performed. If migrating to other network traffic datasets, what adaptive adjustments need to be made to ensure deployment effectiveness on SmartNIC? |
|
ICACT20260019 Question.8
Questioner: zhaozx@dsp.ac.cn 2026-02-09 오후 3:42:42 |
ICACT20260019 Answer.8
Answer by Auhor a.rahman@mst.edu 2026-02-09 오후 3:42:42
Chrome Click!! |
| Is it necessary to accurately constrain the calculation time of intrusion detection algorithms on smart network cards? If the algorithm consumes too much time, will it have a negative impact on the accuracy of intrusion detection？ |
|
ICACT20260019 Question.9
Questioner: md02anas02@gmail.com 2026-02-10 오전 3:52:31 |
ICACT20260019 Answer.9
Answer by Auhor a.rahman@mst.edu 2026-02-10 오전 3:52:31
Chrome Click!! |
| Besides NSL-KDD, do you plan to validate these per-sample latency results using a more modern dataset like CIC-IDS2017? |
|
