adversarial-recommender-systems-survey

本调查的目标有双重：（i）展示对抗性机器学习（ML）以提高RS安全性的最新进展（即，攻击和防御推荐模型），（ii）展示了ML在生成性应用程序的生成对抗网络（GAN）中的另一个成功应用，这要归功于它们学习（多维）数据分布的能力。在这项调查中，我们对主要RS和ML期刊和会议上发表的74篇文章进行了详尽的文献回顾。本评论可供RS社区参考，研究RS的安全性或使用GAN的生成模型以提高其质量。

  • 默认分支master
  • 大小208 KB
  • 更新2021年3月3日
在 GitHub 打开
avatarsisinflab
165
322

技术解读

以下内容由 GitRead 中文社区 编辑团队基于项目 README 与官方资料整理， 结合中文开发者的常见使用场景补充技术选型建议。具体以项目官方文档为准。

深度技术分析：sisinflab/adversarial-recommender-systems-survey

项目背景与痛点

在推荐系统（Recommendation Systems, RS）领域，安全性逐渐成为了新的挑战。推荐系统作为推送个性化内容的关键技术，容易受到恶意攻击，例如操纵推荐结果、注入虚假内容等。为应对这些安全问题，学术界和工业界投入大量资源，研究对抗性机器学习（Adversarial Machine Learning, AML）方法，以提升推荐系统的安全性和鲁棒性。

同时，在生成模型方面，生成对抗网络（Generative Adversarial Networks, GANs）近年来发展迅速，尤其在处理高维度数据时展现了强大的能力。sisinflab/adversarial-recommender-systems-survey这个项目重点探讨了AML技术在推荐系统安全和GANs生成应用中的最新进展。该项目汇总了74篇来自顶级RS和ML期刊与会议的文章，为研究者和开发者提供了详细的文献回顾和前沿技术综述。

核心技术揭秘

该项目从多个角度深入分析了推荐系统的安全性和生成模型的应用。我们从技术架构的角度，来探讨其如何实现对上述技术的研究与应用。

技术实现分析

从技术层面上看，这个项目旨在总结并推广AML在推荐系统中的应用，包括但不限于：

  1. 攻击与防御模型：深入讨论了对抗性攻击策略和防御机制在推荐系统中的应用。研究了许多方法如何提高推荐模型对攻击的抵抗能力。
  2. GANs在高维度数据上的应用：展示了如何利用GANs解决推荐系统中的数据生成问题，特别是在处理复杂的用户行为和内容时。

以下是一个简化的技术架构图，用以说明AML在推荐系统中的核心流程：

graph TD subgraph "推荐系统安全架构" A[用户请求] --> B!("参数校验") B -->|"通过"| C!("执行推荐") B -->|"失败"| D[返回警告] C --> E{"AML 处理"} E -->|"防御不当"| F[返回警告] E -->|"防御成功"| G[结果优化] G --> H[个性化推荐] end

在图中，初始的用户请求通过参数校验后，将执行推荐逻辑。在执行推荐逻辑的过程中，涉及防御性的 AML 处理，以应对攻击。在成功防御之后，推荐结果将被进一步优化，进而生成个性化的推荐结果。

功能亮点与差异

这些设计特点将项目与其他同行区分开来，并展示出其在推荐系统安全领域独特的技术优势。

  • 全面的文献汇总：项目详细归纳了74篇相关文献，形成了详尽的文献综述，为研究者提供了宝贵的资源。
  • 广泛的实施策略：除了讨论防御机制，项目还调查了攻击策略，为理解和解决潜在安全漏洞提供了全面的视角。

应用场景与落地建议

在实际生产环境中应用此项目时，开发者们需要关注以下几个方面：

  1. 安全性考量：在面对推荐系统可能面临的各类攻击时，应采用多种防御策略，并定期更新防御机制。确保系统能适应新的威胁和攻击维度的变化。
  2. 高维度数据处理：利用 GANs 处理高维度数据时，需考虑模型的复杂度和计算资源的需求。精确的参数调整和模型优化显得尤为重要。

综合评价

优点：

  • 提供了丰富的文献参考，方便研究人员进一步挖掘前沿技术。
  • 结合GANs技术，为推荐系统中的生成模型应用提供了新的研究方向。
  • 通过AML引入了关于对抗性攻击与防御的方法论，提高了系统的安全性与鲁棒性。

缺点：

  • 作为综述项目，具体的实现细节和实例较少，可能需要进一步的研究和开发来填补技术鸿沟。
  • 研究过程中不涉及实验代码或工具，实际落地时需要具备相关的技术基础和技术实现能力。

内容已同步仓库最新状态（最后提交：2021年3月3日）。 编辑团队负责选题与抽样审核，详见 编辑准则。

相关仓库

codecrafters-io/build-your-own-x avatar
build-your-own-xcodecrafters-io
7月15日
560.6k
51.6k
289

通过从零复刻你最喜欢的技术来精通编程。

awesome-listfreeprogramming+3
Markdown
1.1 MB
10
详情
sindresorhus/awesome avatar
awesomesindresorhus
9月2日
532.8k
38.2k
103

关于各种有趣主题的精选列表 [注意：在我有机会处理现有拉取请求之前，拉取请求暂时禁用]

awesomeawesome-listlists+2
未知
1.5 MB
1
详情
public-apis/public-apis avatar
public-apispublic-apis
1 天前
496.8k
54.1k
64

免费 API 汇总列表

apipublic-apisfree+7
Python
19.5 MB
7
详情
freeCodeCamp/freeCodeCamp avatar
freeCodeCampfreeCodeCamp
17 小时前
458.7k
48.1k
182

freeCodeCamp.org 的开源代码库和课程体系。免费学习数学、编程和计算机科学。

learn-to-codenonprofitsprogramming+7
TypeScript
571.2 MB
8
详情
morluto/rea avatar
reamorluto
2 小时前
439.1k
108.8k
1.6k

借助智能体逆向分析任何内容，从应用行为直至原生二进制文件。

agent-skillsaiandroid+17
TypeScript
140.6 MB
0
详情
EbookFoundation/free-programming-books avatar
free-programming-booksEbookFoundation
6 天前
401.2k
67.3k
103

免费可获取的编程书籍

bookseducationhacktoberfest+2
Python
21.7 MB
3
详情
openclaw/openclaw avatar
openclawopenclaw
2 小时前
394.1k
82.8k
7.6k

真正做实事的 AI。任何 OS。任何平台。龙虾之道。🦞

aiassistantcrustacean+4
TypeScript
12.8 GB
18
详情
donnemartin/system-design-primer avatar
system-design-primerdonnemartin
9月15日
377.2k
59.2k
637

学习如何设计大规模系统。为系统设计面试做准备。包含 Anki 闪卡。

designdesign-patternsdesign-system+10
Python
11.0 MB
0
详情

关于

GitHub 原始描述（英文）

The goal of this survey is two-fold: (i) to present recent advances on adversarial machine learning (AML) for the security of RS (i.e., attacking and defense recommendation models), (ii) to show another successful application of AML in generative adversarial networks (GANs) for generative applications, thanks to their ability for learning (high-dimensional) data distributions. In this survey, we provide an exhaustive literature review of 74 articles published in major RS and ML journals and conferences. This review serves as a reference for the RS community, working on the security of RS or on generative models using GANs to improve their quality.

描述取自 GitHub 官方 API，保持仓库作者书写的原文。

简要分析

热度分
238
价值分
109
活跃状态
活跃
adversarial-machine-learningfuture-visionsliterature-reviewrecommender-systemadversarial-attackssecuritymachine-learningdeep-learninggangenerative-adversarial-network
GitHub

贡献者

merrafelicemerrafelicetommasodinoiatommasodinoialiuzrccliuzrcc

README

A survey on Adversarial Recommender Systems: from Attack/Defense strategies to Generative Adversarial Networks

A table of adversarial learning publications in recommender systems. This page will be periodically updated to include recent works. Please contact us if your work is not in the list. Let us know if your recent work is not in the list, we will be happy to include it!

The table is complement of the survey below.

A survey on Adversarial Recommender Systems: from Attack/Defense strategies to Generative Adversarial Networks 

@article{DBLP:journals/corr/abs-2005-10322,
    author = {Yashar Deldjoo and Tommaso {Di Noia} and Felice Antonio Merra},
    title = "A survey on Adversarial Recommender Systems: from Attack/Defense strategies to Generative Adversarial Networks",
    journal = "ACM Computing Surveys",
    year = "2021",
    keywords = "Recommender System, Adversarial Machine Learning,
    Literature Review",
    url       = {https://doi.org/10.1145/3439729},
    doi      = {10.1145/3439729}
}

Adversarial Machine Learning in Recommender Systems (AML-RecSys) Tutorial presented at WSDM@2020 slides

@inproceedings{DBLP:conf/wsdm/DeldjooNM20,
  author    = {Yashar Deldjoo and
               Tommaso Di Noia and
               Felice Antonio Merra},
  title     = {Adversarial Machine Learning in Recommender Systems (AML-RecSys)},
  booktitle = {{WSDM} '20: The Thirteenth {ACM} International Conference on Web Search
               and Data Mining, Houston, TX, USA, February 3-7, 2020},
  year      = {2020},
  url       = {https://doi.org/10.1145/3336191.3371877}
  }

Papers

ADVERSARIAL MACHINE LEARNING FOR SECURITY OF RS

Year Title Type Target Model Venue Link Code
2021 Adversarial Item Promotion: Vulnerabilities at the Core of Top-N Recommenders that Use Images to Address Cold Start Attack DVBPR/VBPR/AMR WWW Link Code
2021 A Black-Box Attack Model for Visually-Aware Recommender Systems Attack VBPR/DeepStyle WSDM Link Code
2020 Assessing Perceptual and Recommendation Mutation of Adversarially-Poisoned Visual Recommenders Attack VBPR/AMR NeurIPS-WS Link Code
2020 Attacking Recommender Systems with Augmented User Profiles Attack CF CIKM Link
2020 Multi-Step Adversarial Perturbations on Recommender Systems Embeddings Attack CF arXIv Link Code
2020 Revisiting Adversarially Learned Injection Attacks Against Recommender Systems Attack CF RecSys Link Code
2020 Adversarial Learning for Recommendation: Applications for Security and Generative Tasks — Concept to Code Tutorial BPR-MF RecSys Link Hands-On
2020 TAaMR: Targeted Adversarial Attack against Multimedia Recommender Systems Attack VBPR/AMR DSN-DSML Link Code
2020 Adversarial Training-Based Mean Bayesian Personalized Ranking for Recommender System Attack/Defense BPR-MF IEEE Access Link
2020 Adversarial Learning to Compare: Self-Attentive Prospective Customer Recommendation in Location based Social Networks Attack/Defense LBSN WSDM Link
2020 Privacy-Aware Recommendation with Private-Attribute Protection using Adversarial Learning GAN Defense Attribute-Protection WSDM Link
2020 Adversarial Machine Learning in Recommender Systems (AML-RecSys) Tutorial WSDM Link
2019 Adversarial Collaborative Auto-encoder for Top-N Recommendation Attack/Defense CDAE IJCNN Link
2019 Adversarial Collaborative Neural Network for Robust Recommendation Attack/Defense CDAE SIGIR Link
2019 Adversarial Training Towards Robust Multimedia Recommender System Attack/Defense VBPR TKDE Link Code
2019 Enhancing the Robustness of Neural Collaborative Filtering Systems Under Malicious Attacks Attack/Defense NCF IEEE T Mutimedia Link
2019 Adversarial tensor factorization for context-aware recommendation Attack/Defense FM RecSys Link
2019 Adversarial attacks on an oblivious recommender GAN Attacks Linear RecSys Link
2019 Adversarial Sampling and Training for Semi-Supervised Information Retrieval Attack/Defense MF WWW Link
2019 Domain adaptation in display advertising: an application for partner cold-start Defense Adv. Reg. Deep Rec. RecSys Link
2019 Adversarial Mahalanobis Distance-based Attentive Song Recommender for Automatic Playlist Continuation Attack//Defense MDR SIGIR Paper Code
2018 Adversarial Personalized Ranking for Recommendation Attack/Defense BPR-MF SIGIR Link Code

ADVERSARIAL LEARNING FOR GAN-BASED RECOMMENDATION

Year Title Rec. Model Venue Link Code
2020 LARA: Attribute-to-feature Adversarial Learning for New-item Recommendation Hybrid WSDM Link
2019 Collaborative Adversarial Autoencoders: An Effective Collaborative Filtering Model Under the GAN Framework Collaborative IEEE Access Link
2019 Collaborative Generative Adversarial Network for Recommendation Systems Collaborative ICDE Link
2019 Convolutional Adversarial Latent Factor Model for Recommender System Collaborative AAAI Link
2019 PD-GAN: Adversarial Learning for Personalized Diversity-Promoting Recommendation Collaborative IJCAI Link
2019 LambdaGAN: Generative Adversarial Nets for Recommendation Task with Lambda Strategy Collaborative IJCNN Link
2019 VAEGAN: A Collaborative Filtering Framework based on Adversarial Variational Autoencoders Collaborative IJCAI Link
2019 RsyGAN: Generative Adversarial Network for Recommender Systems Collaborative IJCNN Link
2019 Adversarial Distillation (Transfer) for Efficient Recommendation with External Knowledge Hybrid TIST Link
2019 Adversarial Training for Review-Based Recommendations SIGIR Link
2019 Enhancing Collaborative Filtering with Generative Augmentation Hybrid KDD Link
2019 APL: Adversarial Pairwise Learning for Recommender Systems Collaborative Expert Syst. Appl. Link Code
2019 Generating Reliable Friends via Adversarial Training to Improve Social Recommendation. Social ICDM Link
2019 Utilizing Generative Adversarial Networks for Recommendation based on Ratings and Reviews Collaborative IJCNN Link
2019 A Minimax Game for Generative and Discriminative Sample Models for Recommendation Hybrid PAKDD Link
2019 Leveraging Long and Short-Term Information in Content-Aware Movie Recommendation via Adversarial Training Time-aware IEEE T CYBERNETICS Link
2019 Generative Adversarial User Model for Reinforcement Learning Based Recommendation System CTR ICML Link Code
2019 Adversarial Point-of-Interest Recommendation. POI WWW Link Code
2019 Deep Adversarial Social Recommendation Social IJCAI Link
2019 Click Feedback-Aware Query Recommendation Using Adversarial Examples Query WWW Link
2019 Scenery-Based Fashion Recommendation with Cross-Domain Geneartive Adverserial Networks Fashion BIGCOMP Link
2019 RecSys-DAN: Discriminative Adversarial Networks for Cross-Domain Recommender Systems Fashion IEEE-TNNLS Link
2019 CnGAN: Generative Adversarial Networks for Cross-network user preference generation for non-overlapped users Cross Domain WWW Link Code
2019 C+GAN: Complementary Fashion Item Recommendation Fashion KDD Link
2019 Rating Augmentation with Generative Adversarial Networks towards Accurate Collaborative Filtering Collaborative WWW Link
2019 Privacy and Fairness in Recommender Systems via Adversarial Training of User Representations Privacy ICPRAM Link
2018 CFGAN: A Generic Collaborative Filtering Framework based on Generative Adversarial Networks Collaborative CIKM Link Code
2018 Adversarial Training of Deep Autoencoders Towards Recommendation Tasks Collaborative IC-NIDC Link
2018 Generative Adversarial Network Based Heterogeneous Bibliographic Net Representation for Personalized Citation Rec Collaborative AAAI Link
2018 GraphGAN: Graph Representation Learning With Generative Adversarial Nets Collaborative AAAI Link Code
2018 A Novel Personalized Citation Recommendation Approach Based on GAN Collaborative ISMIS Link
2018 Leveraging Reconstructive Profiles of Users and Items for Tag-Aware Recommendation Hybrid ICDM Link
2018 Rating Prediction in Review-Based Recommendations via Adversarial Auto-Encoder. Hybrid WI Link
2018 PLASTIC: Prioritize Long and Short-term Information in Top-n Recommendation using Adversarial Training. Sequence-aware IJCAI Link
2018 Using Adversarial Autoencoders for Multi-Modal Automatic Playlist Continuation Sequence-aware RecSys Link Code
2018 Multi-Modal Adversarial Autoencoders for Recommendations of Citations and Subject Labels. Sequence-aware UMAP Link Code
2018 Neural Memory Streaming Recommender Networks with Adversarial Training. Sequence-aware KDD Link
2018 RecGAN: recurrent generative adversarial networks for recommendation systems Sequence-aware RecSys Link
2018 Compatibility Family Learning for Item Recommendation and Generation Fashion AAAI Link Code
2018 CRAFT: Complementary Recommendation by Adversarial Feature Transform Fashion ECCV Link
2018 An Adversarial Approach to Improve Long-Tail Performance in Neural Collaborative Filtering Collaborative CIKM Link
2017 Augmented variational autoencoders for collaborative filtering with auxiliary information Collaborative CIKM Link
2017 Visually-Aware Fashion Recommendation and Design with Generative Image Models Fashion ICDM Link Code
2017 IRGAN: A Minimax Game for Unifying Generative and Discriminative Information Retrieval Models Collaborative SIGIR Link Code

Authors

This page is managed and maintained by:

评论

暂无评论