Research map / Label-efficient and robust learning / Recognition and long-tail
Deep Mixture of Diverse Experts for Large-Scale Visual Recognition
Tianyi Zhao, Qiuyu Chen, Zhenzhong Kuang, Jun Yu, Wei Zhang, Jianping Fan B
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 mixture of diverse experts instead of enlarging AlexNet FC8 | AlexNet · ImageNet10K | top-1 accuracy: 31.70% -> 38.65% | Improved |
| add tree-guided task assignment instead of random assignment (ablation) | deep mixture · ImageNet10K | top-1 accuracy: 34.53% -> 38.65% | Improved |
| use two-layer ontology instead of visual tree for task groups (ablation) | deep mixture · ImageNet10K | top-1 accuracy: 37.55% -> 38.65% | Improved |
| add deep multi-task learning within task groups (ablation) | base CNNs · ImageNet10K task groups | top-1 accuracy: 42.3 -> 43.6 (group 1) | Improved |
| increase inter-group overlap lambda from 0.0 to 0.5 (ablation) | deep mixture · ImageNet10K | top-1 accuracy: 33.96% -> 38.65% | Improved |
| increase inter-group overlap lambda from 0.0 to 0.5 (ablation) | deep mixture · ImageNet10K | top-10 accuracy: 57.31% -> 64.32% | Improved |
| use Eq. 8 stacking with not-in-group class instead of Stack 2 (ablation) | deep mixture · ImageNet10K | top-1 accuracy: 37.63% -> 38.65% | Improved |
| late fusion at softmax instead of early fusion at FC7 (ablation) | deep mixture · ImageNet10K | top-1 accuracy: 36.23% -> 38.65% | Improved |
| late fusion at softmax instead of early fusion at FC7 (ablation) | deep mixture · ImageNet10K | top-10 accuracy: 61.38% -> 64.32% | Improved |
Closest papers by meaning
- Multi-Task Dense Prediction via Mixture of Low-Rank Experts (CVPR 2024 · similarity 0.85)
- Multilinear Mixture of Experts: Scalable Expert Specialization through Factorization (NeurIPS 2024 · similarity 0.84)
- Efficient Data Driven Mixture-of-Expert Extraction from Trained Networks (CVPR 2025 · similarity 0.84)
- More Experts Than Galaxies: Conditionally-Overlapping Experts with Biologically-Inspired Fixed Routing (ICLR 2025 · similarity 0.83)
- Unveiling the Unseen: Identifiable Clusters in Trained Depthwise Convolutional Kernels (ICLR 2024 · similarity 0.83)
- Mixture of Experts as Representation Learner for Deep Multi-View Clustering (AAAI 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.
