Research map / Probabilistic and scientific modeling / Variational and generative sampling
Resilient Multiple Choice Learning: A learned scoring scheme with application to audio scene analysis
Victor Letzelter, Mathieu Fontaine, Mickael Chen, Patrick Perez, Slim Essid, Gaël Richard
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 |
|---|---|---|---|
| replace WTA with learned scoring scheme (rMCL), 5 hypotheses | WTA, 5 hyp. · ANSYN D1 | EMD: 48.22 -> 7.04 | Improved |
| replace WTA with rMCL, 5 hypotheses | WTA, 5 hyp. · ANSYN D3 | EMD: 41.83 -> 20.76 | Improved |
| replace WTA with rMCL, 5 hypotheses | WTA, 5 hyp. · ANSYN D3 | Oracle error: 10.44 -> 11.29 | Hurt |
| replace PIT variant with rMCL | PIT variant · RESYN D3 | EMD: 35.05 -> 32.28 | Improved |
| replace PIT variant with rMCL | PIT variant · RESYN D1 | EMD: 10.53 -> 12.14 | Hurt |
| replace PIT variant with rMCL | PIT variant · ANSYN D3 | Oracle error: 18.10 -> 11.29 | Improved |
| use independent ensemble of 5 single-hypothesis models instead of one WTA 1 hyp. | WTA, 1 hyp. · ANSYN D2 | EMD: 24.69 -> 21.64 | Improved |
| use 5 hypotheses instead of 1 for WTA | WTA · ANSYN D1 | EMD: 3.97 -> 48.22 | Hurt |
| add top-n variant on top of rMCL (ablation) | rMCL, 5 hyp. · ANSYN EMD D1-D3 | EMD: 7.04 -> 5.46 | Mixed |
| add epsilon-relaxation to rMCL (ablation) | rMCL, 5 hyp. · ANSYN D2 | EMD: 13.87 -> 12.13 | Improved |
| add epsilon-relaxation to rMCL (ablation) | rMCL, 5 hyp. · ANSYN D3 | Oracle error: 11.29 -> 14.47 | Hurt |
| add epsilon-relaxation to rMCL (ablation) | rMCL, 5 hyp. · RESYN D1 | EMD: 12.14 -> 8.64 | Improved |
Closest papers by meaning
- Mixtures of Experts for Audio-Visual Learning (NeurIPS 2024 · similarity 0.83)
- Multiple Hypothesis Dropout: Estimating the Parameters of Multi-Modal Output Distributions (AAAI 2024 · similarity 0.83)
- Seeing Speech and Sound: Distinguishing and Locating Audio Sources in Visual Scenes (CVPR 2025 · similarity 0.83)
- Improving Multimodal Learning Balance and Sufficiency through Data Remixing (ICML 2025 · similarity 0.83)
- Robust Audio-Visual Segmentation via Audio-Guided Visual Convergent Alignment (CVPR 2025 · similarity 0.83)
- $\texttt{AVROBUSTBENCH}$: Benchmarking the Robustness of Audio-Visual Recognition Models at Test-Time (NeurIPS 2025 · similarity 0.83)
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.
