feat: add TimesNet_Q and xPatch models with Q matrix transformations

2025-08-06 18:39:26 +08:00
parent 7fdf0f364d
commit 6bba6613c9
14 changed files with 872 additions and 3 deletions
--- a/models/xPatch_Q/init.py
+++ b/models/xPatch_Q/init.py
@ -0,0 +1 @@
+from .xPatch_Q import Model
--- a/models/xPatch_Q/network.py
+++ b/models/xPatch_Q/network.py
@ -0,0 +1,132 @@
+import torch
+from torch import nn
+
+class Network(nn.Module):
+    def __init__(self, seq_len, pred_len, patch_len, stride, padding_patch):
+        super(Network, self).__init__()
+
+        # Parameters
+        self.pred_len = pred_len
+
+        # Non-linear Stream
+        # Patching
+        self.patch_len = patch_len
+        self.stride = stride
+        self.padding_patch = padding_patch
+        self.dim = patch_len * patch_len
+        self.patch_num = (seq_len - patch_len)//stride + 1
+        if padding_patch == 'end': # can be modified to general case
+            self.padding_patch_layer = nn.ReplicationPad1d((0, stride)) 
+            self.patch_num += 1
+
+        # Patch Embedding
+        self.fc1 = nn.Linear(patch_len, self.dim)
+        self.gelu1 = nn.GELU()
+        self.bn1 = nn.BatchNorm1d(self.patch_num)
+        
+        # CNN Depthwise
+        self.conv1 = nn.Conv1d(self.patch_num, self.patch_num,
+                               patch_len, patch_len, groups=self.patch_num)
+        self.gelu2 = nn.GELU()
+        self.bn2 = nn.BatchNorm1d(self.patch_num)
+
+        # Residual Stream
+        self.fc2 = nn.Linear(self.dim, patch_len)
+
+        # CNN Pointwise
+        self.conv2 = nn.Conv1d(self.patch_num, self.patch_num, 1, 1)
+        self.gelu3 = nn.GELU()
+        self.bn3 = nn.BatchNorm1d(self.patch_num)
+
+        # Flatten Head
+        self.flatten1 = nn.Flatten(start_dim=-2)
+        self.fc3 = nn.Linear(self.patch_num * patch_len, pred_len * 2)
+        self.gelu4 = nn.GELU()
+        self.fc4 = nn.Linear(pred_len * 2, pred_len)
+
+        # Linear Stream
+        # MLP
+        self.fc5 = nn.Linear(seq_len, pred_len * 4)
+        self.avgpool1 = nn.AvgPool1d(kernel_size=2)
+        self.ln1 = nn.LayerNorm(pred_len * 2)
+
+        self.fc6 = nn.Linear(pred_len * 2, pred_len)
+        self.avgpool2 = nn.AvgPool1d(kernel_size=2)
+        self.ln2 = nn.LayerNorm(pred_len // 2)
+
+        self.fc7 = nn.Linear(pred_len // 2, pred_len)
+
+        # Streams Concatination
+        self.fc8 = nn.Linear(pred_len * 2, pred_len)
+
+    def forward(self, s, t):
+        # x: [Batch, Input, Channel]
+        # s - seasonality
+        # t - trend
+        
+        s = s.permute(0,2,1) # to [Batch, Channel, Input]
+        t = t.permute(0,2,1) # to [Batch, Channel, Input]
+        
+        # Channel split for channel independence
+        B = s.shape[0] # Batch size
+        C = s.shape[1] # Channel size
+        I = s.shape[2] # Input size
+        s = torch.reshape(s, (B*C, I)) # [Batch and Channel, Input]
+        t = torch.reshape(t, (B*C, I)) # [Batch and Channel, Input]
+
+        # Non-linear Stream
+        # Patching
+        if self.padding_patch == 'end':
+            s = self.padding_patch_layer(s)
+        s = s.unfold(dimension=-1, size=self.patch_len, step=self.stride)
+        # s: [Batch and Channel, Patch_num, Patch_len]
+        
+        # Patch Embedding
+        s = self.fc1(s)
+        s = self.gelu1(s)
+        s = self.bn1(s)
+
+        res = s
+
+        # CNN Depthwise
+        s = self.conv1(s)
+        s = self.gelu2(s)
+        s = self.bn2(s)
+
+        # Residual Stream
+        res = self.fc2(res)
+        s = s + res
+
+        # CNN Pointwise
+        s = self.conv2(s)
+        s = self.gelu3(s)
+        s = self.bn3(s)
+
+        # Flatten Head
+        s = self.flatten1(s)
+        s = self.fc3(s)
+        s = self.gelu4(s)
+        s = self.fc4(s)
+
+        # Linear Stream
+        # MLP
+        t = self.fc5(t)
+        t = self.avgpool1(t)
+        t = self.ln1(t)
+
+        t = self.fc6(t)
+        t = self.avgpool2(t)
+        t = self.ln2(t)
+
+        t = self.fc7(t)
+
+        # Streams Concatination
+        x = torch.cat((s, t), dim=1)
+        x = self.fc8(x)
+
+        # Channel concatination
+        x = torch.reshape(x, (B, C, self.pred_len)) # [Batch, Channel, Output]
+
+        x = x.permute(0,2,1) # to [Batch, Output, Channel]
+
+        return x
--- a/models/xPatch_Q/xPatch_Q.py
+++ b/models/xPatch_Q/xPatch_Q.py
@ -0,0 +1,146 @@
+import torch
+import torch.nn as nn
+import math
+import numpy as np
+import os
+
+from layers.decomp import DECOMP
+from .network import Network
+from layers.revin import RevIN
+
+class Model(nn.Module):
+    """
+    xPatch with Q matrix transformation
+    - Applies RevIN normalization first
+    - Applies input Q matrix transformation after RevIN normalization (based on dataset and seq_len)
+    - Uses original xPatch logic (decomposition + dual stream network)
+    - Applies output Q matrix transformation before RevIN denormalization (based on dataset and pred_len)
+    """
+    
+    def __init__(self, configs):
+        super(Model, self).__init__()
+
+        # Parameters
+        seq_len = configs.seq_len   # lookback window L
+        pred_len = configs.pred_len # prediction length (96, 192, 336, 720)
+        c_in = configs.enc_in       # input channels
+
+        # Patching
+        patch_len = configs.patch_len
+        stride = configs.stride
+        padding_patch = configs.padding_patch
+
+        # Store for Q matrix transformations
+        self.seq_len = seq_len
+        self.pred_len = pred_len
+        
+        # Load Q matrices
+        self.load_Q_matrices(configs)
+
+        # Normalization
+        self.revin = configs.revin
+        self.revin_layer = RevIN(c_in, affine=True, subtract_last=False)
+
+        # Moving Average
+        self.ma_type = configs.ma_type
+        alpha = configs.alpha       # smoothing factor for EMA (Exponential Moving Average)
+        beta = configs.beta         # smoothing factor for DEMA (Double Exponential Moving Average)
+
+        self.decomp = DECOMP(self.ma_type, alpha, beta)
+        self.net = Network(seq_len, pred_len, patch_len, stride, padding_patch)
+
+    def load_Q_matrices(self, configs):
+        """Load pre-computed Q matrices for input and output transformations"""
+        # Get dataset name from configs, default to ETTm1 if not specified
+        dataset_name = getattr(configs, 'dataset', 'ETTm1')
+        
+        # Input Q matrix (seq_len)
+        input_q_path = f'cov_mats/{dataset_name}/{dataset_name}_{configs.seq_len}_ratio1.0.npy'
+        # Output Q matrix (pred_len) 
+        output_q_path = f'cov_mats/{dataset_name}/{dataset_name}_{configs.pred_len}_ratio1.0.npy'
+        
+        device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
+        
+        if os.path.exists(input_q_path):
+            Q_input = np.load(input_q_path)
+            self.register_buffer('Q_input', torch.FloatTensor(Q_input).to(device))
+            print(f"Loaded input Q matrix from {input_q_path}, shape: {Q_input.shape}")
+        else:
+            print(f"Warning: Input Q matrix not found at {input_q_path}, using identity matrix")
+            self.register_buffer('Q_input', torch.eye(configs.seq_len).to(device))
+            
+        if os.path.exists(output_q_path):
+            Q_output = np.load(output_q_path)
+            self.register_buffer('Q_output', torch.FloatTensor(Q_output).to(device))
+            print(f"Loaded output Q matrix from {output_q_path}, shape: {Q_output.shape}")
+        else:
+            print(f"Warning: Output Q matrix not found at {output_q_path}, using identity matrix")
+            self.register_buffer('Q_output', torch.eye(configs.pred_len).to(device))
+
+    def apply_input_Q_transformation(self, x):
+        """
+        Apply input Q matrix transformation after RevIN normalization
+        Input: x with shape [B, T, N] where T = seq_len
+        Output: transformed x with shape [B, T, N]
+        """
+        B, T, N = x.size()
+        assert T == self.seq_len, f"Expected seq_len {self.seq_len}, got {T}"
+        
+        # Transpose to [B, N, T] for matrix multiplication
+        x_transposed = x.transpose(-1, -2)  # [B, N, T]
+        
+        # Apply input Q transformation: einsum 'bnt,tv->bnv'
+        # x_transposed: [B, N, T], Q_input.T: [T, T] -> result: [B, N, T]
+        x_trans = torch.einsum('bnt,tv->bnv', x_transposed, self.Q_input.transpose(-1, -2))
+        
+        # Transpose back to [B, T, N]
+        x_transformed = x_trans.transpose(-1, -2)  # [B, T, N]
+        
+        return x_transformed
+
+    def apply_output_Q_transformation(self, x):
+        """
+        Apply output Q matrix transformation to prediction output
+        Input: x with shape [B, pred_len, N]
+        Output: transformed x with shape [B, pred_len, N]
+        """
+        B, T, N = x.size()
+        assert T == self.pred_len, f"Expected pred_len {self.pred_len}, got {T}"
+        
+        # Transpose to [B, N, T] for matrix multiplication
+        x_transposed = x.transpose(-1, -2)  # [B, N, pred_len]
+        
+        # Apply output Q transformation: einsum 'bnt,tv->bnv'
+        # x_transposed: [B, N, pred_len], Q_output: [pred_len, pred_len] -> result: [B, N, pred_len]
+        x_trans = torch.einsum('bnt,tv->bnv', x_transposed, self.Q_output)
+        
+        # Transpose back to [B, pred_len, N]
+        x_transformed = x_trans.transpose(-1, -2)  # [B, pred_len, N]
+        
+        return x_transformed
+
+    def forward(self, x):
+        # x: [Batch, Input, Channel]
+
+        # RevIN Normalization
+        if self.revin:
+            x = self.revin_layer(x, 'norm')
+
+        # Apply input Q matrix transformation after RevIN normalization
+        x_transformed = self.apply_input_Q_transformation(x)
+
+        # xPatch processing with Q-transformed input
+        if self.ma_type == 'reg':   # If no decomposition, directly pass the input to the network
+            output = self.net(x_transformed, x_transformed)
+        else:
+            seasonal_init, trend_init = self.decomp(x_transformed)
+            output = self.net(seasonal_init, trend_init)
+
+        # Apply output Q matrix transformation to the prediction
+        output_transformed = self.apply_output_Q_transformation(output)
+
+        # RevIN Denormalization
+        if self.revin:
+            output_transformed = self.revin_layer(output_transformed, 'denorm')
+
+        return output_transformed