机器学习论文阅读笔记摘录，主动学习Active Learning，迁移学习Transfer Learning，差分隐私Differential Privacy，模型攻击Model Attack
笔记导引
-
- 【Active Learning】
- 【Transfer Learning】
- 【Differential Privacy】
- 【Model inversion attack】
【Active Learning】
【论文笔记01】Learning Loss for Active Learning, CVPR 2019 让网络自己学会预测这个数据的训练损失，从而在无标记Pool中找到valuable instances
【论文笔记02】Active Learning For Convolutional Neural Networks: A Core-Set Approch, ICLR 2018 考虑Diversity的经典approach
【论文笔记03】Variational Adversarial Active Learning, ICCV 2019
【论文笔记04】Ranked Batch-Mode Active Learning, ICCV 2016
主动学习的文章有好多都没防，明年有时间的时候好好更新。
【Transfer Learning】
【论文笔记05】Active Transfer Learning, IEEE T CIRC SYST VID 2020
【论文笔记06】Domain-Adversarial Training of Neural Networks, JMLR 2016 经典DANN
【论文笔记10】A unified framework of active transfer learning for cross-system recommendation, AI 2017
【论文笔记14】Transfer Learning via Minimizing the Performance Gap Between Domains, NIPS 2019 提出了包括最小化Performance Gap在内的四点Principles->GapBoost
【Differential Privacy】
【论文笔记07】A Survey on Differentially Private Machine Learning, IEEE CIM 2020
【论文笔记09】Differentially Private Hypothesis Transfer Learning, ECML&PKDD 2018
【论文笔记11】 Deep Domain Adaptation With Differential Privay, IEEE TIFS 2020 使用类似DANN的网络对抗架构
【论文笔记12coming】Differential privacy based on importance weighting, Mach Learn 2013
【论文笔记13】Differentially Private Optimal Transport: Application to Domain Adaptation, IJCAI 2019 先进行满足差分隐私的OT构造，然后再导出DA框架
【论文笔记15】Boosting and Differential Privacy, 对查询集进行boosting以达到对数据的差分隐私, IEEE ASFCS 2010
【论文笔记16】An Efficient Differential Privacy Logistic Classification Mechanism, 2019 IEEE IoT-J 对输入数据进行干扰
【论文笔记17coming】Semi-Supervised Knowledge Transfer for Deep Learning from Private Training Data, 2017 ICLR
学生在n个teacher的有噪音的投票结果上学习。这篇文章对于 ( ϵ , δ ) (\epsilon, \delta) (ϵ,δ)在实际数据和运算过程中给出比较详细的上界，并在实验中进行比较。
【论文笔记18coming】The Composition Theorem for Differential Privacy
【论文笔记19】Differentially Private Empirical Risk Minimization, 2011 JMLR
(General frameworks of output perturbation, objective perturbation, and parameter fine-tuning, including the specific generalization bound for linear regression, SVMs! This paper’s theory has a lot of constraints on the loss function and the regularizer)
【论文笔记20】Functional Mechanism Regression Analysis under Differential Privacy, 2012 VLDB
(linear regression and logistic regression. The key of using objective perturbation is to have a polynomial form/approximation of the cost function)
【论文笔记21】Differential Privacy Preservation for Deep Auto-Encoders: An Application of Human Behavior Prediction, 2016 AAAI
【论文笔记22】Differentially Private Empirical Risk Minimization Revisited-Faster and More General, 2017 NIPS
【Model inversion attack】
【论文笔记08】Model inversion attacks that exploit confidence information and basic countermeasures, SIGSAC 2015