ICACT20210223 Question.1
Questioner: Prof. Babu Baniya 2021-02-09 오후 12:56:25 |
ICACT20210223 Answer.1
Answer by Auhor yfhou@okayama-u.ac.jp 2021-02-09 오후 12:56:25
Chrome  Click!! |
| Thank you for sharing an interesting article in IEEE/ICACT2021, can you please tell us QoS of your proposed method? |
Thank you for your question. In our proposed method, we choose throughput as the major QoS for prediction because many applications depend on such value to decide whether the wireless application can be employed for usage. Of course, some other parameters of QoS can also be used for the proposed method. |
ICACT20210223 Question.2
Questioner: ch.lee@etri.re.kr 2021-02-10 오후 12:51:51 |
ICACT20210223 Answer.2
Answer by Auhor yfhou@okayama-u.ac.jp 2021-02-10 오후 12:51:51
Chrome Click!! |
| Thank you for your presentation. Do you have any special reasons for using CNN rather than LSTM? |
Thank you for your good question.It is ture that LSTM is often used for the prediction of time-series data. Our proposal majorly considers using CNN to reduce the computation complexity of data training since CNN can reduce the data size by adjusting the size of filter |
ICACT20210223 Question.3
Questioner: franklinlu888@outlook.com 2021-02-10 오후 12:54:38 |
ICACT20210223 Answer.3
Answer by Auhor yfhou@okayama-u.ac.jp 2021-02-10 오후 12:54:38
Chrome Click!! |
| Do you think the Convolutional Neural Network can be improved in your study? How? |
Thank you for your question. The CNN can obtain promising performance especially when the targets of input data or image have an even distribution. However, when target includes a few values which occupy major distribution, the prediction process is apt to make the predicted value near to these values after training process. Therefore, some rare values are neglected which may reduce the prediction accuracy. This is major reason that we make the vectorization for the prediction target to change its distribution for good prediction accuracy. |
