Research map / Detection and segmentation / CNN features
Deep Transfer Learning for Multiple Class Novelty Detection
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 |
|---|---|---|---|
| Add membership loss to cross-entropy training (ablation) | Single CNN · Caltech-256 | AUC: 0.854 -> 0.865 | Helped |
| Add a parallel reference-data branch with cross-entropy training (ablation) | CNN novelty detector · Caltech-256 | AUC: 0.854 -> 0.864 | Helped |
| Combine membership loss and parallel reference-data training (ablation) | CNN novelty detector · Caltech-256 | AUC: 0.854 -> 0.906 | Helped |
| Use Caltech-256 rather than ILSVRC12 as the reference dataset (ablation) | Deep Novelty · CUB-200 | AUC: 0.958 -> 0.953 | Hurt |
| Use Dogs rather than ILSVRC12 as the reference dataset (ablation) | Deep Novelty · CUB-200 | AUC: 0.958 -> 0.945 | Hurt |
| Use a larger, more diverse reference dataset (ablation) | Deep Novelty · CUB-200 | AUC: 0.945 -> 0.958 | Helped |
| Apply membership loss and globally negative filter training | VGG16 fine-tuning · Caltech-256 | classification accuracy: 0.908 -> 0.939 | Helped |
| Apply membership loss and globally negative filter training | VGG16 fine-tuning · CUB-200 | classification accuracy: 0.988 -> 0.990 | Helped |
| Apply membership loss and globally negative filter training | VGG16 fine-tuning · Dogs | classification accuracy: 0.730 -> 0.801 | Helped |
| Apply membership loss and globally negative filter training | VGG16 fine-tuning · FounderType | classification accuracy: 0.945 -> 0.950 | Helped |
Closest papers by meaning
- Hierarchical Novelty Detection for Visual Object Recognition (CVPR 2018 · similarity 0.91)
- Adversarially Learned One-Class Classifier for Novelty Detection (CVPR 2018 · similarity 0.89)
- Where's Wally Now? Deep Generative and Discriminative Embeddings for Novelty Detection (CVPR 2019 · similarity 0.88)
- Learning Deep Classifiers Consistent With Fine-Grained Novelty Detection (CVPR 2021 · similarity 0.88)
- CSI: Novelty Detection via Contrastive Learning on Distributionally Shifted Instances (NeurIPS 2020 · similarity 0.87)
- Universal Novelty Detection Through Adaptive Contrastive Learning (CVPR 2024 · 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.