目录 Introduction
Why semi-supervised learning help?
Semi-supervised Learning for Generative Model
Supervised Generative Model
Semi-supervised Generative Model
Low-density Separation Assumption
Self Training
Entropy-based Regularization(基…
SupCon 定义:
Clusters of points belonging to the same class are pulled together in embedding space, while simultaneously pushing apart clusters of samples from different classes.
novelties:
属于同一类的归一化后的特征表示靠得越近越好…
Weakly Supervised Semantic Segmentation by Pixel-to-Prototype Contrast
CVPR2022 北京大学 Though image-level weakly supervised semantic segmentation (WSSS) has achieved great progress with Class Activation Maps (CAMs) as the cornerstone, the large supervis…
supervised pca很简单粗暴,计算 X X X的每一个纬度和 Y Y Y的相关性,取一个阈值,丢掉一些纬度,然后用普通的pca降维。 如何计算两个随机变量的相关性/相似性? 两个随机变量 X , Y X,Y X,Y,有一个函数 ϕ \phi ϕ,可以把一维随机变量映射到高维空间,映射后两个向量均值的…