Code release for Proto-CLIP: Vision-Language Prototypical Network for Few-Shot Learning | IROS 2024
-
Updated
Dec 30, 2024 - Python
Code release for Proto-CLIP: Vision-Language Prototypical Network for Few-Shot Learning | IROS 2024
A Lightweight and High Performance Neural network for MI-EEG decoding
Hyperbolic Busemann Learning with Ideal Prototypes, NeurIPS2021
Official Pytorch Code of Our Paper: Rethinking Multiple Instance Learning for Whole Slide Image Classification: A Good Instance Classifier is All You Need
[MM2024] PyTorch implementation of "Semantic Codebook Learning for Dynamic Recommendation Models".
[CIKM 2024] PRISM: Mitigating EHR Data Sparsity via Learning from Missing Feature Calibrated Prototype Patient Representations
A demo showing the benefits of using HDC in a telecom incident scenario
Prototype-guided contrastive learning for long-tailed steel surface defect classification. PyTorch implementation of our paper under review at Electronics Measurement Technology (北大核心).
Official PyTorch implementation of Ternary Relational Networks (TRN) & Context-Gated Signed Evidence Units.
Source code for the ADAM submission on the PlantCLEF2025 challenge
Which prototype aggregation rules resist collapse in continual crop-disease learning: HDC vs NCM, RanPAC-style, Sign/Median-NCM on DINOv3 features
Official project page for ProME; public data access now, research code after manuscript acceptance.
Explainable Few-Shot Learning for Medical Image Diagnosis using Prototype Networks, OOD Detection and Grad-CAM Visualization
Official PROTON+ implementation: contamination-robust prototype banks for test-time online OOD detection with medical VLMs. Follow-up to PROTON (MICCAI 2026). Live browser demo, benchmarks, evaluation toolkit.
Frozen encoder + Mahalanobis prototype for class-incremental intent classification. 50+ experiments across BANKING77, CLINC150, HWU64, AG News. Matches fine-tuned baselines at 5MB state with zero forgetting, order-invariance, 455 QPS.
Noisy-label fine-grained visual recognition with OpenAI CLIP ViT-B/32, four-view TTA & Prototype-Consistency Robust Adapter (PCR-Adapter). Runs end-to-end on Apple Silicon (MPS).
A lightweight Few-Shot Image Classifier built with PyTorch, capable of recognizing new classes from just a few examples using prototype-based learning. Designed for easy model export and integration.
How a bird classifier changes its evidence when parts move. Two-minute presentation, prototype scores, and reproducible Shapley–Taylor analysis.
🚀 GPU-accelerated prototype learning framework. Condenses infinite data streams into differentiable topological grids, reducing metric inference complexity from O(N×D) to strictly O(K×D) .
To associate your repository with the prototype-learning topic, visit your repo's landing page and select "manage topics."