【2017_ICCV_DML]Smart Mining for Deep Metric Learning
来源:互联网 发布:车铣复合手工编程例子 编辑:程序博客网 时间:2024/06/17 08:28
- 为了解决DML问题得到feature embedding,目前主要用triplet model减少类内差异,增大类间差异,但是大量的训练样本会导致收敛很慢。这个问题促进了embedding的global structure发展和hard negative|positive,可是这些方法通常计算量比较大。文章提出的方法结合了Triplet model和embedding space的global structure,通过smart mining过程产生有效的训练样本,除此之外,还提出了一个adaptive controller能够自动调整smart mining 超参数。
- DML通常面临的场景是类别特别多,每类的样本特别少。训练过程中通常需要选择hard triplet,这样能够获得有效大量级的梯度,加速收敛。
阅读全文
0 0
- 【2017_ICCV_DML]Smart Mining for Deep Metric Learning
- Deep Metric Learning for Person Re-Identification
- metric deep learning loss
- papers for metric learning
- 【2017_ICCV】Deep Metric Learning with Angular Loss
- Discriminative Deep Metric Learning for Face Verification in the Wild(文献泛读)
- 泛读:CVPR2014:Discriminative Deep Metric Learning for Face Verification in theWild
- 【论文笔记】Margin Sample Mining Loss: A Deep Learning Based Method for Person Re-identification
- 【Person Re-ID】Margin Sample Mining Loss: A Deep Learning Based Method for Person Re-identification
- A Matlab Toolkit for Distance Metric Learning
- Similarity Metric Learning for Face Recognition2013
- Optimization for Deep Learning Highlights in 2017
- Deep Learning for Beginners
- Deep Learning for OCR
- 阅读小结Deep Metric Learning via Lifted Structured Feature Embedding
- 【Deep Learning】笔记:Tips for deep learning
- Neural Networks (Deep Learning) , NLP and Text Mining
- 行人检索 - Embedding Deep Metric for Person Re-identification
- 基于SignalR的小型IM系统
- centos7安装mysql5.6
- 认识Java
- Thinking in java之构造器
- Reactjs入门官方文档(三)【components-and-props】
- 【2017_ICCV_DML]Smart Mining for Deep Metric Learning
- python函数语法学习
- Redis五种数据类型介绍
- sunny-ngrok 的配置及使用
- Intellij IDEA常用快捷键大全
- Hi3516A-常用指令和根文件目录详解
- 浅谈深度学习
- ZQ0001-《清单革命》
- 欢迎使用CSDN-markdown编辑器