FilterBank

class dpeeg.transforms.FilterBank(freq: float, filter_bank: list = [[4, 8], [8, 12], [12, 16], [16, 20], [20, 24], [24, 28], [28, 32], [32, 36], [36, 40]], transition_bandwidth: float = 2.0, gstop: float = 30, gpass: float = 3)

Filter Bank.

EEG data will be filtered according to different filtering frequencies and finally concatenated together. eg.`(Batch, …) -> (Batch, F, …)` if the number of filter banks exceeds 1, (Batch, …) -> (Batch, …) if the filter has only one. By default, filtering is performed on edata, please ensure the availability of the data. Related references include [1] and [2].

Parameters:

• freq (float) – EEG data sampling frequency.

• filter_bank (multiple 2 float of list) – The low-pass and high-pass cutoff frequencies for each filter set.

• transition_bandwidth (float) – The bandwidth (in hertz) of the transition region of the frequency response from the passband to the stopband.

• gstop (float) – The minimum attenuation in the stopband (dB).

• gpass (float) – The maximum loss in the passband (dB).

Returns:

data – Transformed eegdata.

Return type:

eegdata or dataset

References

[1]

R. Mane, E. Chew, K. Chua, K. K. Ang, N. Robinson, A. P. Vinod, S.-W. Lee, and C. Guan, “FBCNet: A multi-view convolutional neural network for brain-computer interface,” arXiv preprint arXiv:2104.01233, 2021.

[2]

X. Ma, W. Chen, Z. Pei, J. Liu, B. Huang, and J. Chen, “A temporal dependency learning CNN with attention mechanism for MI-EEG decoding,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2023.

Examples

>>> eegdata = dpeeg.EEGData(edata=np.random.randn(16, 3, 10),
...                         label=np.random.randint(0, 3, 16))
>>> transforms.FilterBank(250)(eegdata, verbose=False)
[edata=(16, 9, 3, 10), label=(16,)]