www.icact.org

Now 6837 visitors
Today:0 Yesterday:0
Total: 155528 216S 51P 5R
2026-10-22, Week 43

Member Login

TACT Journal Homepage

Call for Paper

Committee

Find My Paper

Author Homepage

Statistics & History

Welcome Message

Paper Procedure

FAQ

Paper Archives

Outstanding Papers

Program & Proceedings

Presentation Platform

Hotel & Travel Info

Photo Gallery

Scheduler Login

Seminar

Archives Login

Sponsors

Online Q&A Session


Paper Number	Author:
Paper Title
Keyword
Q&A Number	**Select your question number !!
Questioner eMail:
Question	Thank you for your talk! I am curious if you plan to explore other datasets to see if you can generalize better the offloading coefficient, if you think that a layered offloading might help at all, and if you tested whether or not the offloading had any negative effects such as stopping overfitting?
Answer by Author	Thank you for your insightful question! Addressing each of your points: Exploring Other Datasets for Generalization: Yes, we recognize the importance of evaluating our offloading strategy across diverse datasets to ensure broader applicability. Currently, our framework is optimized for specific scenarios, but we plan to test it on datasets from other domains, such as industrial IoT and smart city applications. This will help refine the offloading coefficient and further enhance generalizability. Layered Offloading: A hierarchical or layered offloading approach is indeed an interesting direction. While our current method relies on uncertainty-based semantic offloading, we are considering a more granular offloading mechanism that dynamically selects multiple levels of RESs or even cloud resources based on real-time network conditions and computational constraints. This could further optimize decision-making, especially in multi-tiered edge-cloud architectures. Impact of Offloading on Overfitting: Offloading in our framework is guided by uncertainty (entropy), meaning that only data samples with high uncertainty are offloaded to higher-capacity models at the RES. This does not inherently cause overfitting, as we are not reusing the same data excessively or memorizing specific patterns. Instead, this approach improves generalization by ensuring that uncertain samples receive more sophisticated processing, reducing the likelihood of local edge models becoming biased toward their own limited data distribution. However, we do acknowledge the importance of systematically analyzing how offloading influences model learning dynamics, and future studies will include detailed evaluations on this aspect. We appreciate your interest in our work and look forward to further refining our approach based on these considerations!

Save Q&A
* Edit or answer any Q&A by selecting Q&A number Hyper Link below + Write button (Save)

ICACT20250118 Question.1
Questioner: taikung@gmail.com 2025-02-18 오후 3:37:10

ICACT20250118 Answer.1
Answer by Auhor avi@khu.ac.kr 2025-02-18 오후 3:37:10
Chrome Click!!

Thank you for your talk! I am curious if you plan to explore other datasets to see if you can generalize better the offloading coefficient, if you think that a layered offloading might help at all, and if you tested whether or not the offloading had any negative effects such as stopping overfitting?

Thank you for your insightful question! Addressing each of your points: Exploring Other Datasets for Generalization: Yes, we recognize the importance of evaluating our offloading strategy across diverse datasets to ensure broader applicability. Currently, our framework is optimized for specific scenarios, but we plan to test it on datasets from other domains, such as industrial IoT and smart city applications. This will help refine the offloading coefficient and further enhance generalizability. Layered Offloading: A hierarchical or layered offloading approach is indeed an interesting direction. While our current method relies on uncertainty-based semantic offloading, we are considering a more granular offloading mechanism that dynamically selects multiple levels of RESs or even cloud resources based on real-time network conditions and computational constraints. This could further optimize decision-making, especially in multi-tiered edge-cloud architectures. Impact of Offloading on Overfitting: Offloading in our framework is guided by uncertainty (entropy), meaning that only data samples with high uncertainty are offloaded to higher-capacity models at the RES. This does not inherently cause overfitting, as we are not reusing the same data excessively or memorizing specific patterns. Instead, this approach improves generalization by ensuring that uncertain samples receive more sophisticated processing, reducing the likelihood of local edge models becoming biased toward their own limited data distribution. However, we do acknowledge the importance of systematically analyzing how offloading influences model learning dynamics, and future studies will include detailed evaluations on this aspect. We appreciate your interest in our work and look forward to further refining our approach based on these considerations!

Now 6837 visitors Today:0 Yesterday:0 Total: 155528 216S 51P 5R 2026-10-22, Week 43

Now 6837 visitors
Today:0 Yesterday:0
Total: 155528 216S 51P 5R
2026-10-22, Week 43