Prototype Memory

Prototype memory is a machine learning approach that uses representative examples, or prototypes, to classify data points, offering advantages in scenarios with limited labeled data or noisy information. Current research focuses on improving prototype generation and selection, often incorporating techniques like optimal transport and quality-aware weighting to enhance accuracy and robustness, particularly within models such as prototype networks and their hierarchical extensions. These advancements are impacting diverse fields, improving performance in tasks ranging from image classification and face recognition to named entity recognition and semantic segmentation in remote sensing. The ability to learn effectively from limited or imperfect data makes prototype memory a valuable tool across numerous applications.

Papers

October 30, 2024

Incremental Learning of Retrievable Skills For Efficient Continual Task Adaptation
Daehee Lee, Minjong Yoo, Woo Kyung Kim, Wonje Choi, Honguk Woo
Imitation Learning Incremental Learning Adaptation Concern Task Adaptation Efficient Continual Learning Skill Transfer Non Cognitive Skill Prototype Memory

January 22, 2024

Augmenting Prototype Network with TransMix for Few-shot Hyperspectral Image Classification
Chun Liu, Longwei Yang, Dongmei Dong, Zheng Li, Wei Yang, Zhigang Han, Jiayao Wang
Hyperspectral Image Spectral Information Spatial Spectral Soft QMIX Prototype Memory Shot Hyperspectral Image Classification

December 18, 2023

Predict and Interpret Health Risk using EHR through Typical Patients
Zhihao Yu, Chaohe Zhang, Yasha Wang, Wen Tang, Jiangtao Wang, Liantao Ma
Deep Learning Model Electronic Health Record Visual Prototype Health Risk Prediction Prototype Memory Progressive Integration Network

November 13, 2023

Quality-Aware Prototype Memory for Face Representation Learning
Evgeny Smirnov, Vasiliy Galyuk, Evgeny Lukyanets
Face Recognition Model Face Embeddings Face Recognition Benchmark Prototype Memory

October 12, 2023

MProto: Multi-Prototype Network with Denoised Optimal Transport for Distantly Supervised Named Entity Recognition
Shuhui Wu, Yongliang Shen, Zeqi Tan, Wenqi Ren, Jietian Guo, Shiliang Pu, Weiming Lu
Entity Recognition Optimal Transport Entity Representation Class Incremental NER Prototype Memory

August 16, 2022

Unsupervised domain adaptation semantic segmentation of high-resolution remote sensing imagery with invariant domain-level prototype memory
Jingru Zhu, Ya Guo, Geng Sun, Libo Yang, Min Deng, Jie Chen
Semantic Segmentation Deep Convolutional Neural Network Unsupervised Domain Adaptation High Resolution Semantic Segmentation Network Visual Imagery Prototype Memory

November 30, 2021

Hierarchical Prototype Networks for Continual Graph Representation Learning
Xikun Zhang, Dongjin Song, Dacheng Tao
Catastrophic Forgetting Graph Representation Learning Topological Information Citation Network Continual Graph Prototype Memory

Prototype Memory

Papers

Incremental Learning of Retrievable Skills For Efficient Continual Task Adaptation

Augmenting Prototype Network with TransMix for Few-shot Hyperspectral Image Classification

Predict and Interpret Health Risk using EHR through Typical Patients

Quality-Aware Prototype Memory for Face Representation Learning

MProto: Multi-Prototype Network with Denoised Optimal Transport for Distantly Supervised Named Entity Recognition

Unsupervised domain adaptation semantic segmentation of high-resolution remote sensing imagery with invariant domain-level prototype memory

Hierarchical Prototype Networks for Continual Graph Representation Learning