Research map / Label-efficient and robust learning / Recognition and long-tail
Reconstructing Randomly Masked Spectra Helps DNNs Identify Discriminant Wavenumbers
Yingying Wu, Jinchao Liu, Yan Wang, Stuart James Gibson, Margarita Osadchy, Yongchun Fang
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 |
|---|---|---|---|
| add joint training of reconstruction and classifier instead of separate training (x10) (ablation) | TeaNet · RRUFF_IR | accuracy (%): 87.96 -> 90.07 | Improved |
| add joint training instead of separate training (x10) (ablation) | TeaNet · USGS | accuracy (%): 84.65 -> 87.37 | Improved |
| add joint training instead of separate training (x10) (ablation) | TeaNet · ReLab | accuracy (%): 83.05 -> 85.43 | Improved |
| increase masked samples from x5 to x10 (joint) (ablation) | TeaNet · RRUFF_IR | accuracy (%): 89.05 -> 90.07 | Improved |
| increase masked samples from x10 to x20 (joint) (ablation) | TeaNet · RRUFF_IR | accuracy (%): 90.07 -> 89.66 | Hurt |
| use random mask ratio U(0.3,0.7) instead of U(0.1,0.9) (ablation) | TeaNet · USGS | accuracy (%): 84.73 -> 87.37 | Improved |
| use random mask position and length (MS4) instead of fixed first half mask (MS1 a=0,b=0.5) (ablation) | TeaNet · ReLab | accuracy (%): 80.89 -> 85.43 | Improved |
| set near-total mask ratio tau=0.99 instead of 0.5 (ablation) | TeaNet · RRUFF_IR | accuracy (%): 90.07 -> 72.18 | Hurt |
| replace masked reconstruction with CutMix rough filling | CNN_Partial · USGS | accuracy (%): 86.59 -> 85.43 | Hurt |
| add masked-spectra training (CNN_Partial) to CNN_Full | CNN_Full · RRUFF_IR | accuracy (%): 88.03 -> 86.53 | Hurt |
| add masked-spectra training (CNN_Partial) to CNN_Full | CNN_Full · ReLab | accuracy (%): 82.06 -> 83.90 | Improved |
| add TeaNet reconstruction augmentation to CNN_Full | CNN_Full · ReLab | accuracy (%): 82.06 -> 85.43 | Improved |
Closest papers by meaning
- Analytical-Chemistry-Informed Transformer for Infrared Spectra Modeling (AAAI 2025 · similarity 0.84)
- RamanBench: A Large-Scale Benchmark for Machine Learning on Raman Spectroscopy (NeurIPS 2026 · similarity 0.81)
- Frequency-Aware Deepfake Detection: Improving Generalizability through Frequency Space Domain Learning (AAAI 2024 · similarity 0.81)
- GLACIER: Rethinking Mass Spectrum Prediction as an Object Detection Problem (NeurIPS 2026 · similarity 0.80)
- Addressing Spectral Bias of Deep Neural Networks by Multi-Grade Deep Learning (NeurIPS 2024 · similarity 0.80)
- Langevin-Informed Transfer Learning: Replacing the Target Samples by Black-Box Feedback (NeurIPS 2026 · 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.
