I am working on the sleep stage classification. I read some research articles about this topic many of them used SVM or ensemble method. Is it a good idea to use convolutional neural network to classify one-dimensional EEG signal? I am new to this kind of work. Pardon me if I ask anything wrong?
There is nothing wrong on using CNN. But the explain-ability matters. ML is enough for these type of problems.
Convolutional learning methods are more efficient to learn local patterns than recurrent neural network in the time series classification tasks.
Intuitively, 1D-CNN are often applied. However, in one of my experiments, 2D-CNN outperforms 1D-CNN after I reshape my 1D time series into a 2D matrix.
Dear Muhammad Ali , Xin Bi , Tao Tao , Manoj Koyadan Thank you very much.
You can use 1D CNN as well for such kind of problems. Note that CNN requires appropriate selection of the network architecture to handle such problems. However, the problem is the computational time. For large number of input and output variables, the algorithm requires rather a long time to converge. In such cases, you can use SVM, subject to obtaining desirable results from the algorithm.
CNN is good for 1D signal, although I would suggest LSTM. You only need to be sure to understand the basic concept of all suitable expert systems for a specific problem as the results are always dependent on the expert's understanding and the setting of the hyper-parameters.
Yes, you can use a CNN for 1D signal or you try a simple multiperceptron neural network to classify the 1D signal.
Thank you Purnendu Mishra Abdulrahaman Otuoze Mokhaled N. A. Al-Hamadani Davood Moghadas
Yes, i have also used CNNs for a similar work and got better results with respect to ML models.
Depending on the problem, you can use CNN for extracting features, or even a combination of CNN with temporal models like GRU or LSTM to better embrace the time element which is an important part of the EEG signal.
Dear Fazla Rabbi Mashrur
This article may help you
Article Novel Feature Selection and Voting Classifier Algorithms for...
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science