Dear community , currently working on emotions recognition , as a first step I'm trying to extract features , I was checking some recources , I found that they used the SEED dataset , it contains EEG signals of 15 subjects that were recorded while the subjects were watching emotional film clips. Each subject is asked to carry out the experiments in 3 sessions. There are 45 experiments in this dataset in total. Different film clips (positive, neutral, and negative emotions) were chosen to receive highest match across participants. The length of each film clip is about 4 minutes. The EEG signals of each subject were recorded as separate files containing the name of the subjects and the date. These files contain a preprocessed, down-sampled, and segmented version of the EEG data. The data was down-sampled to 200 Hz. A bandpass frequency filter from 0–75 Hz was used. The EEG segments associated with every movie were extracted. There are a total of 45 .mat files, one for each experiment. Every person carried out the experiment three times within a week. Every subject file includes 16 arrays; 15 arrays include preprocessed and segmented EEG data of 15 trials in one experiment. An array named LABELS contains the label of the corresponding emotion- al labels (−1 for negative, 0 for neutral, and +1 for positive). I found that they loaded each dataset separately (negative , neutral , positive) , and they fixed length of signal at 4096 and number of signal for each class at 100 , and fixed number of features extracted from Wavelet packet decomposition at 83 , my question is why they selected 83 , 4096 and 100 exactly ?
I know that my question is a bit long but I tried to explain clearly the situation , I appreciate your help thank you .
Length of signal 4096 is perhaps due to the sampling rate. You can check this from the description of the database. How many samples per second for EEG data? Usually, for EEG the sampling rate is not to high. Somewhere between 200 to 500 Hz. From sampling rate you can know the time duration of the signal. If sampling rate is 256 Hz for instance, then the duration of your signal would be 4096/256 = 16 seconds.Then for computation purpose, the length of data is chosen in power of two such as 32, 64, 128, 256,512, 1024, 2048 .. and that's why there is 4096 .. if for example the data length is 4000, if you use Matlab to calculate the FFT, it will zero pad the data so that the length would become to the nearest higher power of two, which in this case is 4096
Ahmad Nasrul Norali Thank you for your answer , what about the selection of 83 features and number of 100 signal per class exactly ?
Hello,
Check these papers for insight
https://www.hindawi.com/journals/js/2018/9709505/
https://www.researchgate.net/publication/4319558_Feature_Extraction_Technique_using_Discrete_Wavelet_Transform_for_Image_Classification
Regards
As far as the problem is concerned, when DWT is executed as a feature extraction method, lots of wavelet coefficients are generated as the decomposition level increases. Therefore, choosing the necessary features from DWT coefficients becomes crucial for the effectiveness and conciseness of the classifier. Generally, this issue can be solved by manual selection of the features. However, to determine the practicability of DWT an active feature selection (ASF) strategy for DWT coefficients can be adapted. There are two phases contained in the ASF process. In phase 1, to reduce and provide a uniform feature dimensionality, we use relative power ratio (RPR) to pre-process each DWT coefficient obtained from digital responses. Further, in phase 2, the optimal DWT coefficients for classification with an automatic search method are identified to ensure that the outcomes of the whole searching process are the most useful features for the classification. Through the above phases, the original DWT features are refined and selected according to their RPR values. Concise features are finally achieved in an active manner with no human designation.
For the number 83 and 100, perhaps it's based on number of features, number of channels, number of wavelet packet etc.
I've looked at the SEED website just to figure it out and looking at the description there is no mention on using wavelet in the project. And also i notice they used "extracted differential entropy (DE) features" and " differential asymmetry (DASM) and rational asymmetry (RASM) features as the differences and ratios between the DE features of 27 pairs of hemispheric asymmetry electrodes ". So the numbers you are asking about, you may get the answer from here.
That's as far as I can tell. But the numbers of features perhaps depends on how many types of features is used? how many channels they used for valid data? If there is some kind of data decomposition, perhaps this might also contribute to some number. So basically there is nothing unusual about the number. It depends on how they tabulate and structured the data and labels.
And the 100 signals, sorry I got no idea because I don't have the data to look at the structure.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics