Research map / Time series / Long-term forecasting
TFD-Net: Towards Intelligent Time-Frequency Mode Decomposition with Practical Applications
Pingping Pan, Yunjian Zhang, Jinyi Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use TFD-Net instead of IVGNMD for mode decomposition | TFD-Net · Simulated test set, SNR 20 dB | Reconstruction MSE: 0.295 -> 0.079 | Improved |
| use TFD-Net instead of ACMD for TF mode separation | TFD-Net · Simulated test set, SNR 5 dB | TF modes DICE loss: 0.885 -> 0.587 | Improved |
| add TFR loss supervision in the TFA module (ablation) | TFD-Net · Simulated test set (SNR sweep) | TFR/TF mode DICE, MSE | Improved |
| add residual connections in the U-Net block (ablation) | TFD-Net · Simulated test set (SNR sweep) | TFR/TF mode DICE, MSE | Improved |
| add enhanced channel attention in the TF transform block (ablation) | TFD-Net · Simulated test set (SNR sweep) | TFR/TF mode DICE, MSE | Improved |
| increase TF transform convolution filters from one to three (sizes 31, 63, 95) (ablation) | TFD-Net · Validation loss | validation loss | Improved |
| increase TF transform filters beyond three (four or five filters) (ablation) | TFD-Net · Validation loss | validation loss | No clear effect |
| increase U-Net convolutional layers from 16 to 24 (ablation) | TFD-Net · Validation loss | validation loss | Improved |
| increase U-Net convolutional layers from 24 to 32 (ablation) | TFD-Net · Validation loss | validation loss | No clear effect |
| apply TFD-Net to real oscillation and echolocation data without retraining | TFD-Net · Real-world signals (qualitative) | mode decomposition quality | Improved |
Closest papers by meaning
- DTF-AT: Decoupled Time-Frequency Audio Transformer for Event Classification (AAAI 2024 · similarity 0.82)
- HyFAD: Hybrid Time-Frequency Diffusion with Frequency-Aware Embedding for Time Series Imputation (NeurIPS 2026 · similarity 0.82)
- Content-aware Balanced Spectrum Encoding in Masked Modeling for Time Series Classification (AAAI 2025 · similarity 0.80)
- TimeMixer++: A General Time Series Pattern Machine for Universal Predictive Analysis (ICLR 2025 · similarity 0.80)
- RTFS-Net: Recurrent Time-Frequency Modelling for Efficient Audio-Visual Speech Separation (ICLR 2024 · similarity 0.80)
- TimeKAN: KAN-based Frequency Decomposition Learning Architecture for Long-term Time Series Forecasting (ICLR 2025 · similarity 0.80)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
