Atlas8346-CSDN博客

原创【论文精读】Learn from Neighbour: A Curriculum that Train Low Weighted Samples By Imitation

Learn from Neighbour: A Curriculum that Train Low Weighted Samples By Imitation原文地址：Learn from Neighbour: A Curriculum that Train Low Weighted Samples By Imitation这篇文章提出了一种新的课程学习的策略（见论文第4节），主要思想就是，一...

2018-10-09 16:13:45 310

原创【论文精读】Select Via Proxy: Efficient Data Selection For Training DeepNetworks

Select Via Proxy: Efficient Data Selection For Training DeepNetworks2019ICLR的文章，介绍了Select Via Proxy（SVP）通过较小规模的模型来确定样本的uncertainty进而决定复杂模型训练使用样本的策略。原文地址：Select Via Proxy: Efficient Data Selection Fo...

2018-10-08 13:18:16 981

原创【论文精读】Improving Simple Models with Confidence Profiles

Improving Simple Models with Confidence Profiles原文地址：Improving Simple Models with Confidence ProfilesAbstract用ProWeight方法进行模型迁移，使用linear probes通过flattened intermediate representations生成confidence s...

2018-10-03 14:54:01 465

原创【论文精读】Learning Bounds for Importance Weighting

Learning Bounds for Importance Weighting原论文地址：Learning Bounds for Importance WeightingAbstract1 Introduction现实世界中机器学习训练数据和测试数据样本的分布会有偏差。一个常见的修正方法叫做importance weighting，它通过给不同的训练样本的带价值赋予权重来平衡这种偏差。一...

2018-10-02 15:01:29 1488

原创【论文精读】Curriculum Learning

Curriculum Learning论文原文：Curriculum Learning课程学习（Curriculum Learning）由Montreal大学的Bengio教授团队在2009年的ICML会议上提出，主要思想是模仿人类学习的特点，由简单到困难来学习课程（在机器学习里就是容易学习的样本和不容易学习的样本），这样容易使模型找到更好的局部最优，同时加快训练的速度。Abstract人...

2018-10-01 08:56:57 28534 1

原创【论文精读】MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labels

MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labels原文地址：MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labels补充证明：Su...

2018-08-10 11:26:49 8127 4

原创【论文精读】Knowledge Transfer with Jacobian Matching

Knowledge Transfer with Jacobian Matching原文链接：Knowledge Transfer with Jacobian Matching

2018-08-09 19:38:29 1473

原创【论文精读】Deep Learning and the Information Bottleneck Principle

Deep Learning and the Information Bottleneck Principle原文链接：Deep Learning and the Information Bottleneck Principle上次读了《Deep Learning under Privileged Information Using Heteroscedastic Dropout》这篇文章，...

2018-07-28 11:15:06 5505 2

原创【论文精读】Dropout: A Simple Way to Prevent Neural Networks from Overfitting

Dropout: A Simple Way to Prevent Neural Networks from Overfitting原文链接：Dropout: A Simple Way to Prevent Neural Networks from Overfitting来填坑了，上次读了《Deep Learning under Privileged Information Using He...

2018-07-26 22:41:29 8392 2

原创【论文精读】Deep Learning under Privileged Information Using Heteroscedastic Dropout

Deep Learning under Privileged Information Using Heteroscedastic Dropout原文链接：Deep Learning under Privileged Information Using Heteroscedastic Dropout模型和代码：Learned models and the source code这篇文章发...

2018-07-21 21:43:20 3397 4

原创【论文精读】3D Shape Segmentation

3D Shape Segmentation with Projective Convolutional Networks原文地址：3D Shape Segmentation with Projective Convolutional Networks数据和代码：Data and codeAbstract这篇文章介绍了一种分割3D物体的深度架构，此架构结合了Fully Convo...

2018-07-18 14:28:36 2357 2

原创【论文精读】Multi-view Convolutional Neural Networks for 3D Shape Recognition

Multi-view Convolutional Neural Networks for 3D Shape Recognition原文地址：Multi-view Convolutional Neural Networks for 3D Shape Recognition 代码和数据：Code and data这篇文章使用三维物体Multi-view（多视角）的二位渲染图片作为训练数据，基...

2018-07-14 11:04:33 7752 1

原创【论文精读】A Support Vector Clustering Method

A Support Vector Clustering Method原文地址：A Support Vector Clustering Method本文介绍了一种支持向量聚类的方法。Abstract本文展示了使用支持向量进行数据聚类的方法，使用核技巧将数据从低维映射到高维，在高维空间中，边界是超球面，低维中边界显示为不规则的几何形状。Describing Cluster ...

2018-07-12 10:52:45 1063

原创【论文精读】SVM for Clustering

A Support Vector Method for Clustering原文地址：A Support Vector Method for Clustering本文介绍了一种基于SVM的聚类方法，核心思想是用高斯核的SVM找到多个能够包围数据的半径最小的超球。使用此方法不用预先确定类别的结构和个数。Abstract聚类问题可以用参数化或非参数化的方法处理。参数化方法往往限制于其表...

2018-07-11 12:02:54 2578 2

原创【李宏毅深度学习】Tips for Training Deep Neural Network

李宏毅深度学习_Tips for Training Deep Neural Network本文是李宏毅深度学习 (2015)的学习笔记，主要介绍了在训练DNN过程中的不同阶段用到的一些技巧。本文所用到的图示主要来自课堂ppt。原视频地址：李宏毅深度学习 (2015)概述想要提高深度学习的效率和收获比较好的结果，可以从以上五部分入手，下面分别从每个部分入手，介绍一些常用于深度...

2018-05-25 10:20:46 984

原创【李宏毅深度学习】Backpropagation

本文是李宏毅深度学习 (2015)的学习笔记，主要介绍了神经网络Backpropagation算法的推导过程。本文所用到的图示均来自课堂ppt。原视频地址：[李宏毅深度学习 (2015)][1]

2018-05-23 12:01:20 831

qq_25011449的博客