Anchor-driven Multi-modal Multi-scale Expert Selection for Survival Prediction
cs.CV
Submitted: 2026-10-06
Updated: 2026-10-06
Code: https://github.com/taozh2017/AM2ES
Sources
- Anchors Aweigh! Sail for Optimal Unified Multi-Modal Representations
- Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer
- Med-MoE: Mixture of Domain-Specific Experts for Lightweight Medical Vision-Language Models
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models