Research map / Label-efficient and robust learning / Recognition and long-tail
SagaNet: A Small Sample Gated Network for Pediatric Cancer Diagnosis
Read the paperCheck 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 mask input plus gated layers, reconstruction loss and length-aware hinge loss (full SagaNet) vs. baseline without these parts (ablation) | SagaNet · SRBCT2 | accuracy: 0.5550 -> 0.8031 | Helped |
| remove mask input (use raw images) (ablation) | SagaNet · SRBCT2 | accuracy: 0.8031 -> 0.7457 | Hurt |
| replace gated layers with vanilla convolutions (ablation) | SagaNet · SRBCT2 | accuracy: 0.8031 -> 0.7058 | Hurt |
| remove reconstruction regularization loss (ablation) | SagaNet · SRBCT2 | accuracy: 0.8031 -> 0.6268 | Hurt |
| replace length-aware hinge loss with cross-entropy (ablation) | SagaNet · SRBCT2 | accuracy: 0.8031 -> 0.6086 | Hurt |
| ImageNet pre-training for comparison CNNs (mask input) | DenseNet/ResNet/MobileNet etc. · SRBCT2 | accuracy: e.g. DenseNet121 0.6150 -> 0.7568 | Helped |
| use mask input instead of raw input for pre-trained baselines | pre-trained comparison CNNs · SRBCT2 | accuracy: DenseNet121pre 0.7568 (mask) vs 0.6793 (raw); ResNet50pre 0.7268 (mask) vs 0.6396 (raw); Inception-v3pre 0.6565 (mask) v | Mixed |
| SagaNet with mask input vs. ImageNet pre-trained DenseNet201 with mask input | SagaNet · SRBCT2 | accuracy: 0.7847 -> 0.8031 | Helped |
| full SagaNet on BreakHist vs classical classifiers | SagaNet · BreakHist 40x | recognition rate: 96.2 vs best classical 83.8 (QDA) | Helped |
Closest papers by meaning
- STARC-9: A Large-scale Dataset for Multi-Class Tissue Classification for CRC Histopathology (NeurIPS 2025 · similarity 0.82)
- Adversarial Learning Under Hybrid Perturbations for Robust Acute Lymphoblastic Leukemia Classification (AAAI 2025 · similarity 0.82)
- Cell Graph Transformer for Nuclei Classification (AAAI 2024 · similarity 0.81)
- NaFV-Net: An Adversarial Four-view Network for Mammogram Classification (AAAI 2025 · similarity 0.81)
- Beyond ViT Tokens: Masked-Diffusion Pretrained Convolutional Pathology Foundation Model for Cell-Level Dense Prediction (NeurIPS 2026 · similarity 0.81)
- ZePT: Zero-Shot Pan-Tumor Segmentation via Query-Disentangling and Self-Prompting (CVPR 2024 · similarity 0.81)
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.