accuracy is lower than presented in paper

[Ericzzzzzzzzz]

i used the files in preprocessing to process the seed dataset and that train the model ,everything is fine however the test accuracy seem to The accuracy converged to around 70%. This much lower then what is presented in the paper，I am wondering why is this happening
did i miss some peperprocess step?

[Ericzzzzzzzzz]

我使用预处理中的文件来处理种子数据集并训练模型，一切都很好，但是测试准确率似乎收敛到 70% 左右。这比论文中提出的要低得多，我想知道为什么会发生这种情况， 我是否错过了一些预处理步骤？

Never mind ,I found my mistakes ,this project is great

[eeyhsong]

Good luck to you research 😉

[dake-sun]

我使用预处理中的文件来处理种子数据集并训练模型，一切都很好，但是测试准确率似乎收敛到 70% 左右。这比论文中提出的要低得多，我想知道为什么会发生这种情况， 我是否错过了一些预处理步骤？

Never mind ,I found my mistakes ,this project is great

I have the same problem as you. I used the way mentioned in the paper for the seed dataset and also trained with the conformer model, but ended up with a very low accuracy on the test dataset, what did you do to correct it?

[Ericzzzzzzzzz]

我使用预处理中的文件来处理种子数据集并训练模型，一切都很好，但是测试准确率似乎收敛到 70% 左右。这比论文中提出的要低得多，我想知道为什么会发生这种情况， 我是否错过了一些预处理步骤？

Never mind ,I found my mistakes ,this project is great

I have the same problem as you. I used the way mentioned in the paper for the seed dataset and also trained with the conformer model, but ended up with a very low accuracy on the test dataset, what did you do to correct it?

In fact,it still bother me that in my conformer model that the in each subject,first fold of the experiment is always much more lower than the other 4 fold for example the subject 1 the first fold is about 76% and the other is 95%,I dont konw why and it not suppose to happend. I used author's frame to divide the dataset and train the model.maybe you should try it. hope it can help you

[dake-sun]

OK, thanks for your advice! I have another question, what is the shape of your input data? Is it (3394, 1, 62, 200)?

[Ericzzzzzzzzz]

OK, thanks for your advice! I have another question, what is the shape of your input data? Is it (3394, 1, 62, 200)?

我的输入是 n，1，62，200，n是多少个样本我没注意过，应该是3394

[dake-sun]

OK, thanks for your advice! I have another question, what is the shape of your input data? Is it (3394, 1, 62, 200)?

我的输入是 n，1，62，200，n是多少个样本我没注意过，应该是3394

好的，感谢