Pytorch的distributed training 现在数据集变得越来越大,训练也由原来的多卡到多机,pytorch在1.0.0之后在distributed learning上做的已经比较好了,在官网也有现成的文档可以看,网上对这部分内容的介绍也很多。这里只是记录下自己看tutorial的过程,可能会有很多不足的地方,欢迎指正。 2020-04-19 Tools #Pytorch
Bit-Scalable Deep Hashing With Regularized Similarity Learning for Image Retrieval and Person Re-Identification TIP2015 paper,除了利用deep learning,为了能够实现变长的hash code,学习了weight matrix来对hamming distance做加权。 2019-11-25 Computer Science # Deep Hash
Supervised Hashing for Image Retrieval via Image Representation Learning AAAI2014论文,第一篇基于Deep Learning的Learning to hash文章,采用的双阶段方法,先把由ground truth得到的similarity matrix进行decomposition,得到近似的target hash code,再学习image到target hash code的映射。 2019-11-22 Computer Science #Deep Hash
Simultaneous Feature Learning and Hash Coding with Deep Neural Networks CVPR2015 deep hash论文,采用triplet loss并提出一个divide-and-encode module。 2019-11-22 Computer Science #Deep Hash
awesome-hash Collections 1. Statistics 2. Survey 3. Unsupervised Hash (+ Transfer / Semi-supervised learning) 4. Supervised Hash 5. Others 6. Datasets 2019-11-20 Computer Science > Collections #hash
最大熵模型的理解 最近在看一些无监督聚类的文章,发现了很有意思的博客,里面讲了很多关于数学、物理、天文还有信息科学的内容。在看完博主关于最大熵的内容后也写下一点心得。 首先是熵的定义,离散概率和连续概率形式:$$S(x)=-\sum_{x} p(x) \log p(x)\tag{1}$$$$S(x)=-\int p(x)\log p(x) dx\tag{2}$$ 其实之间看到这个定义是比较疑惑的,为什么会是这个形式 2019-11-08 Math #概率论 #最大熵 #熵
Unsupervised Cross-dataset Person Re-identification by Transfer Learning of Spatial-Temporal Pattern CVPR2018 paper,ReID上的transfer learning。 2019-10-21 Computer Science #Unsupervised ReID #Transfer Learning
Self-similarity Grouping: A Simple Unsupervised Cross Domain Adaptation Approach for Person Re-identification ICCV 2019 oral 2019-09-11 Computer Science #ReID #Unsupervised #Grouping
STA: Spatial-Temporal Attention for Large-Scale Video-based Person Re-Identification AAAI2019 paper 2019-09-10 Computer Science #Video Person ReID #Attention
Cross-view Asymmetric Metric Learning for Unsupervised Person Re-identification ICCV2017 paper 2019-09-10 Computer Science #ReID #Unsupervised #Cross-view #Metric Learning