Research map / Label-efficient and robust learning / OOD and anomaly detection
Towards an Incremental Unified Multimodal Anomaly Detection: Augmenting Multimodal Denoising From an Information Bottleneck Perspective
Kaifang Long, Lianbo Ma, Jiaqi Liu, Liming Liu, Guoyang Xie
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 Mamba decoder (ablation) | IB-IUMAD · MVTec 3D-AD RGB+3D, 6-1 with 4 steps | I-AUROC: 75.3 -> 76.0 | Improved |
| add IBFM (ablation) | IB-IUMAD · MVTec 3D-AD RGB+3D, 6-1 with 4 steps | I-AUROC: 75.3 -> 76.9 | Improved |
| add both Mamba decoder and IBFM (ablation) | IB-IUMAD · MVTec 3D-AD RGB+3D, 6-1 with 4 steps | I-AUROC: 75.3 -> 78.6 | Improved |
| add both Mamba decoder and IBFM (ablation) | IB-IUMAD · MVTec 3D-AD RGB+3D, 6-1 with 4 steps | forgetting metric: 12.7 -> 9.3 | Improved |
| add both Mamba decoder and IBFM (ablation) | IB-IUMAD · MVTec 3D-AD RGB+3D, 9-1 with 1 step | I-AUROC: 85.1 -> 87.5 | Improved |
| add both Mamba decoder and IBFM (ablation) | IB-IUMAD · MVTec 3D-AD RGB+3D, 10-0 with 0 step | I-AUROC: 86.7 -> 91.0 | Improved |
| use cross-attention fusion instead of addition (ablation) | IB-IUMAD · MVTec 3D-AD RGB+3D, 10-0 with 0 step | I-AUROC: 89.8 -> 91.0 | Improved |
| use cross-attention fusion instead of LinearGLU (ablation) | IB-IUMAD · MVTec 3D-AD RGB+3D, 10-0 with 0 step | I-AUROC: 86.7 -> 91.0 | Improved |
| IB-IUMAD vs IUF in 6-1 with 4 steps (RGB+3D) | IUF · MVTec 3D-AD RGB+3D, 6-1 with 4 steps | I-AUROC: 75.1 -> 78.6 | Improved |
| IB-IUMAD vs CDAD in 9-1 with 1 step (RGB) | CDAD · MVTec 3D-AD RGB, 9-1 with 1 step | I-AUROC: 74.2 -> 85.3 | Improved |
| IB-IUMAD vs CDAD on Eyecandies 3D forgetting, 9-1 with 1 step | CDAD · Eyecandies 3D, 9-1 with 1 step | forgetting metric: 0.9 / 0.1 -> 1.4 / 1.3 | Hurt |
| IB-IUMAD vs CDAD on Eyecandies RGB+3D, 10-0 with 0 step | CDAD · Eyecandies RGB+3D, 10-0 with 0 step | I-AUROC: 79.2 -> 80.6 | Improved |
Closest papers by meaning
- Towards Zero-Shot Anomaly Detection and Reasoning with Multimodal Large Language Models (CVPR 2025 · similarity 0.86)
- Revisiting Multimodal Fusion for 3D Anomaly Detection from an Architectural Perspective (AAAI 2025 · similarity 0.85)
- Multimodal Industrial Anomaly Detection by Crossmodal Feature Mapping (CVPR 2024 · similarity 0.85)
- DZAD: Diffusion-based Zero-shot Anomaly Detection (AAAI 2025 · similarity 0.85)
- MMDiff: Multimodal Model Diffing for Feature Discovery and Control (NeurIPS 2026 · similarity 0.85)
- CNC: Cross-modal Normality Constraint for Unsupervised Multi-class Anomaly Detection (AAAI 2025 · similarity 0.85)
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.
