You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am doing the same things as you in recent time. Have you written your papers yet? If I hope to give me a link, I will read it carefully.
I have read three papers about this dataset include Oshea, but all the accuracy they said I think is impossible. if you have studied the dataset carefully,You will find that all deep learning methods have no effect on QAM16 and QAM64. I specially training networks to classify only those two types of signals, and the accuracy of the verification set is always around 50%, both CNN and RNN has been tried, and many other research. i'm totally sure that this question won't be solved by DL, unless we change the data a bit.
But in this three papers their minimum accuracy reachs 87.4% and the highest is more than 91%, this is just a joke that they use the simplest model to get the unbelievable accuracy.
Your results are very close to mine, indicating that our study is without any false information, we hope that we can work together to solve the problem of QAM signal classification.
The text was updated successfully, but these errors were encountered:
I am doing the same things as you in recent time. Have you written your papers yet? If I hope to give me a link, I will read it carefully.
I have read three papers about this dataset include Oshea, but all the accuracy they said I think is impossible. if you have studied the dataset carefully,You will find that all deep learning methods have no effect on QAM16 and QAM64. I specially training networks to classify only those two types of signals, and the accuracy of the verification set is always around 50%, both CNN and RNN has been tried, and many other research. i'm totally sure that this question won't be solved by DL, unless we change the data a bit.
But in this three papers their minimum accuracy reachs 87.4% and the highest is more than 91%, this is just a joke that they use the simplest model to get the unbelievable accuracy.
Your results are very close to mine, indicating that our study is without any false information, we hope that we can work together to solve the problem of QAM signal classification.
The text was updated successfully, but these errors were encountered: