Frequency-domain MLPs are More Effective Learners in Time Series Forecasting.
Kun YiQi ZhangWei FanShoujin WangPengyang WangHui HeNing AnDefu LianLongbing CaoZhendong NiuPublished in: NeurIPS (2023)
Keyphrases
- frequency domain
- spatial domain
- fourier transform
- power spectrum
- denoising
- frequency analysis
- fourier domain
- subband
- fourier analysis
- cross correlation
- feature extraction
- low pass
- facial asymmetry
- neural network
- spectrum analysis
- power spectra
- bandpass
- mathematical formalism
- gabor filters
- multilayer perceptron
- non stationary
- wavelet transform
- discrete fourier transform
- high quality
- frequency response
- frequency spectrum