![](https://img-blog.csdnimg.cn/20201014180756925.png?x-oss-process=image/resize,m_fixed,h_64,w_64)
论文笔记
文章平均质量分 84
MC.zeeyoung
这个作者很懒,什么都没留下…
展开
-
【论文笔记】Long Tail Learning via Logit Adjustment
重温了基于标签频率的逻辑调整,并将这种logit adjustment应用在训练前还是还是训练中。这种调整是支持原始的logit和主类之间的相对边界。原创 2023-03-01 14:44:27 · 1098 阅读 · 0 评论 -
【论文笔记】Distribution Alignment: A Unified Framework for Long-tail Visual Recognition
we develop an adaptive calibration function that enables us to adjust the classification scores for each data point. We then introduce a generalized re-weight method in the two-stage learning to balance the class prior, which provides a flexible and unifie原创 2023-03-01 11:56:55 · 241 阅读 · 0 评论 -
【论文笔记】Decoupling Representation and Classifier for Long-Tailed Recognition
现状:Existing solutions usually involve class-balancing strategies, e.g. by loss re-weighting, data re-sampling, or transfer learning from head- to tail-classes, but most of them adhere to the scheme of jointly learning representations and classifiers.原创 2023-02-28 17:17:09 · 449 阅读 · 0 评论 -
【论文阅读】Revisiting Long-tailed Image Classification: Survey and Benchmarks with New Evaluation Metrics
【基于当前许多真实世界场景中的数据分布都呈现长尾分布,因此,许多关于长尾分布的图像分类的成果被许多人所关注】【然而,他们通常评估一个平衡测试集或多个独立测试集的性能,都需要这些测试集具有与训练数据不同的分布。【当前测试数据可能存在相似的分布,现存评估测量无法客观反映出模型的泛化性能】原创 2023-02-20 16:14:19 · 490 阅读 · 1 评论 -
【论文阅读】Automated detection of Alzheimer disease using MRI images and deep neural networks- A review
Early detection of Alzheimer’s disease is crucial for deploying interventions and slowing the disease progression. The purpose of this survey is to provide an overview of recent research on deep learning models for Alzheimer’s disease diagnosis.The first原创 2023-02-08 00:47:26 · 130 阅读 · 1 评论 -
论文笔记 “General Multi-label Image Classification with Transformers”
摘要多标签图像分类是一个预测一张图中若干目标、属性以及其他实体所对应的标签语义。本文提出Classification Transformer (C-Trans) 网络来实现通用的多标签图像分类任务。这个网络利用Transformer来探索视觉特征和标签之间的复杂依赖关系。主要流程是,首先利用Transformer对图像进行特征编码,从而预测出三个东西,一个是目标标签集合,另一个是给定的masked 标签,以及卷积网络中所得到的视觉特征。在这三个东西中最重要的是对训练目标进行标签的mask。这个mask是以原创 2023-02-01 01:16:26 · 784 阅读 · 2 评论 -
【论文笔记】Weakly supervised multiple instance learning histopathological tumor segmentation
这篇文章太过抽象,如果有纰漏之处,还望指出 ,不胜感激。这篇文章是属于多实例学习(MIL)的一个病理组织学方面的应用创新。创新贡献点在于:1)从WSI(whole slide image)二值生成标签,用来训练分割模型2)通过只训练二值WSI标注来生成实例级别的预测模型3)公开可用的肿瘤标注库TCGA已经发布,含6481个样本比之前的多出一个数量级4)WSI预处理工具的开源讲完这里,我...原创 2020-05-01 20:43:16 · 2165 阅读 · 0 评论