Series decomposition block

Multilayer perceptron to encode/decode high dimension representation of sequential data

Special designed layernorm for the seasonal part

AutoCorrelation Mechanism with the following two phases: (1) period-based dependencies discovery (2) time delay aggregation This bloc

De-stationary Attention

Performs a 1d DWT Forward decomposition of an image Args: J (int): Number of levels of decomposition wave (str or pywt.Wavelet o

Performs a 1d DWT Inverse reconstruction of an image Args: wave (str or pywt.Wavelet or tuple(ndarray)): Which wavelet to use.

Autoformer encoder

Autoformer decoder layer with the progressive decomposition architecture

Autoformer encoder

Autoformer encoder layer with the progressive decomposition architecture

1D multiwavelet block.

MLP to learn the De-stationary factors Paper link: https://openreview.net/pdf?id=ucNDIDRNjjv

The Two Stage Attention (TSA) Layer input/output shape: [batch_size, Data_dim(D), Seg_num(L), d_model]

Bottleneck convolution CSCM

Series decomposition block

DataEmbedding with configurable kernel size(`d_kernel`) and sequence length(`seq_len`). To solve the warning for EigenWorms dataset (seq_len=1798

A encoder model with self attention mechanism.

Compose with two layers

Fourier Filter: to time-variant and time-invariant term

A demonstration of finding one step transition of linear system by DMD iteratively

Find koopman transition of linear system by DMD with multistep K approximation

LayerNorm but with an optional bias. PyTorch doesn't support simply bias=False

Smart Lazy-Loading Dictionary

MIC layer to extract local and global features

Mamba Block with support for time-variant dt, B, C. The time-variant parameters are controlled by `timevariant_dt`, `timevariant_B`, and `tim