# One-Shot-Learning

**Repository Path**: AuspicL/One-Shot-Learning

## Basic Information

- **Project Name**: One-Shot-Learning
- **Description**: 小样本学习的一些方法
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 1
- **Forks**: 0
- **Created**: 2021-09-14
- **Last Updated**: 2024-10-04

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# One-Shot-Learning
小样本学习的一些方法<br>
这大概是2019年3-4月两周看的相关方法。<br>

**One shot learning**是一种对象分类问题，出现于计算机视觉领域。尽管大多数基于机器学习的对象分类算法需要对数百或数千个样本/图像和非常大的数据集进行训练，但是一次性学习旨在从一个或仅少数训练样本/图像中学习关于对象类别的信息。<br>
**Few shot learning** 同样是用少量训练样本训练模型。<br>
**Zero Shot Learning** (零样本学习) 属于迁移学习的一个分支.在 Zero Shot Learning 中训练集和测试集的类别无交集。比如训练集中有马，老虎，测试时要预测斑马。<br>

2006LiFeiFei:One-Shot Learning of Object Categories <br>
2013HBPL:One-shot learning by inverting a compositional causal process <br>
度量学习：根据不同的任务来自主学习出针对某个特定任务的度量距离函数。 <br>
度量学习CVPR2018：Learning to Compare: Relation Network for Few-Shot Learning <br>
孪生网络Siamese neural networks for one-shot image recognition(2015) <br>
匹配网络Matching Networks for One Shot Learning(2016) <br>
原型网络2017Prototypical networks for few-shot learning <br>
基于图2018Few-shot learning with graph neural networks <br>
递归记忆模型2016Meta-Learning with Memory-Augmented Neural Networks <br>
优化器学习2017Optimization as a model for few-shot learning <br>
模型无关自适应2017Model-agnostic meta-learning for fast adaptation of deep networks <br>