Research map / Architectures and efficiency / Neural architecture search
MFAS: Multimodal Fusion Architecture Search
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 MFAS rather than random search to select fusion architectures | Multimodal fusion search on AV-MNIST · AV-MNIST validation | accuracy: 0.9203 (0.0021) -> 0.9264 (0.0004) | Helped |
| Search the fusion architecture instead of using CentralNet | AV-MNIST · AV-MNIST test | accuracy (%): 87.86 -> 88.38 | Helped |
| Search the fusion architecture instead of using CentralNet | MM-IMDB · MM-IMDB | weighted F1: 0.6223 -> 0.6250 | Helped |
| Search the fusion architecture instead of using GMU | MM-IMDB · MM-IMDB | macro F1: 0.5410 -> 0.5568 | Helped |
| Search the fusion architecture instead of using CentralNet | NTU RGB+D · NTU RGB+D cross-subject | accuracy (%): 89.36 -> 90.04 | Helped |
| Replace concatenation with weighted-sum fusion (ablation) | MFAS fusion layers · Multimodal classification experiments | classification performance | Hurt |
| Use 1D convolution instead of fully connected fusion layers (ablation) | MFAS fusion layers · Multimodal classification experiments | classification performance | No clear effect |
Closest papers by meaning
- MFH-NAS:A Hybrid Neural Architecture Search Framework for Multimodal Fusion Object Detection (ICML 2026 · similarity 0.89)
- Fusion or Confusion? Multimodal Complexity Is Not All You Need (NeurIPS 2026 · similarity 0.87)
- DC-NAS: Divide-and-Conquer Neural Architecture Search for Multi-Modal Classification (AAAI 2024 · similarity 0.86)
- OSKAR: Omnimodal Self-supervised Knowledge Abstraction and Representation (NeurIPS 2025 · similarity 0.86)
- Beyond Modality Fusion: Deep Ensembles for Multimodal Classification (NeurIPS 2026 · similarity 0.85)
- Are We Making Progress in Multimodal Domain Generalization? A Comprehensive Benchmark Study (NeurIPS 2026 · similarity 0.84)
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.