Hi, all! I'm a Research Scientist at Netflix, where I works on the combination of large language model and generative recommendation. My goal? To enhance your viewing experiences with better, personalized video recommendations. I graduated from Purdue Computer Science. I am fortunate to be supervised by Prof. Xiangyu Zhang and co-advised by Prof. Jean Honorio. My PhD research focuses on Robust Machine Learning. Before coming to Purdue, I received my B.S. from National Elite Program at Nanjing University and was supervised by Prof. Yang Yu. I also collaborated with Xin Lu from Adobe and Pannaga Shivaswamy from Netflix.

/ /

Last updated at Jan 2024

News


  • [06/2024] We got 1 paper accepted in KDD-EAI 2024.
  • [01/2024] We got 1 paper accepted in AAAI 2024.
  • [08/2023] I will join Netflix as a Research Scientist!
  • [05/2023] I started summer internship in Google Research!
  • [01/2023] Our paper on robust backdoor removal is accepted in CVPR 2023
  • [11/2022] We got 1 paper accepted in ICLR 2023
  • [08/2022] Our paper on safety of federated learning got Best Paper Award in ECCV 2022 AROW Workshop.
  • [05/2022] I start research internship in Netflix Research.
  • [05/2022] We got one paper on backdoor defense accepted by ICML 2022.
  • [03/2022] We got two papers accepted by CVPR 2022, congrats on all the collaboraters.
  • [12/2021] Our paper on input inversion is accepted by NDSS 2022
  • [10/2021] Our paper on backdoor removal is accepted by S&P 2022
  • [05/2021] I start summer research internship at Adobe Emerging Product Group!
  • [05/2021] Our paper on trojan model detection is accepted by ICML 2021.
  • [04/2021] Our paper on deriving robustness upper bound through information theory is accepted by ISIT 2021.
  • [12/2020] Our paper on feature space attack is accepted by AAAI 2021.
  • [11/2020] My advisor Xiangyu Zhang, is honored with title Samuel D. Conte.
  • [01/2020] Our paper on debugging neural network is accepted by ICSE 2020.
  • [10/2019] My undergrad work on end-to-end combination of reasoning and perception is accepted by NeurIPS 2019.

Selected Publications [Full List]

(* represents equal contribution)

Responsive image

POSIT: Promotion of Semantic Item Tail via Adversarial Learning
Qiuling Xu , Pannaga Shivaswamy, and Xiangyu Zhang
IEEE Transactions on Knowledge Discovery and Data Mining (KDD 2024) Workshop
[Arxiv] [Bibtex] [Code] [Workshop]


Responsive image

MEDIC: Remove Model Backdoors via Importance Driven Cloning
Qiuling Xu , Guanhong Tao, Jean Honorio, Yinqi Liu, Shengwei An, Guangyu Shen, Siyuan Cheng, and Xiangyu Zhang
IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2023)
[Paper] [Bibtex] [Code] [Talk]


Responsive image

FLIP: A Provable Defense Framework for Backdoor Mitigation in Federated Learning
Kaiyuan Zhang, Guanhong Tao, Qiuling Xu , Siyuan Cheng, Yinqi Liu, Shengwei An, Guangyu Shen, and Xiangyu Zhang
International Conference on Learning Representations (ICLR 2023)
[Paper]


Responsive image

Reducing Accuracy Gap in Adversarial Training by Discriminating Adversarial Samples
Qiuling Xu , Shengwei An, Guanhong Tao, Jean Honorio and Xiangyu Zhang
Technical Report
[Manuscript]


Responsive image

Better Trigger Inversion Optimization in Backdoor Scanning
Guanhong Tao , Guangyu Shen , Yingqi Liu , Shengwei An , Qiuling Xu , Shiqing Ma , Pan Li , and Xiangyu Zhang
IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2022) (Oral)
[Paper] [Bibtex]


Responsive image

Bounded Adversarial Attack on Deep Content Features
(STOA trade-off between imperceptibility and success)
Qiuling Xu , Guanhong Tao and Xiangyu Zhang
IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2022)
[Paper] [Code] [Bibtex] [Talk] [Poster] [Arxiv]


Responsive image

Constrained Optimization with Dynamic Bound-scaling for Effective NLP Backdoor Defense
Guangyu Shen , Yingqi Liu , Guanhong Tao Qiuling Xu , Zhuo Zhang , Shiqing Ma , Shengwei An and Xiangyu Zhang
International Conference on Machine Learning (ICML 2022)
[Paper] [Bibtex]


Responsive image

Model Orthogonalization: Class Distance Hardening in Neural Networks for Better Security
Guanhong Tao , Yingqi Liu , Guangyu Shen , Qiuling Xu , Shengwei An , Zhang Zhuo , Pan Li , and Xiangyu Zhang
Proceedings of IEEE Symposiums on Security and Privacy (S&P 2022)
[Paper] [Bibtex]


Responsive image

A Le Cam Type Bound for Adversarial Learning and Applications
Qiuling Xu* , Kevin Bello*, and Jean Honorio
2021 IEEE International Symposium on Information Theory (ISIT 2021)
[Arxiv] [Bibtex] [Short Talk] [Long Talk]


Responsive image

Towards Feature Space Adversarial Attack by Style Perturbation
Qiuling Xu , Guanhong Tao , Siyuan Cheng and Xiangyu Zhang
35th AAAI Conference on Artificial Intelligence (AAAI 2021)
[Arxiv] [Code] [Bibtex] [Short Talk] [Long Talk] [Poster]


Responsive image

Bridging Machine Learning and Logical Reasoning by Abductive Learning
Wangzhou Dai*, Qiuling Xu*, Yang Yu* and Zhihua Zhou
32th Advances in Neural Information Processing Systems (NeurIPS 2019)
[Paper] [Code] [Arxiv] [Bibtex]

Honors


Services


Conference Reviewer: ICLR 2023, AAAI 2023, NeurIPS 2022, CVPR2022, ICML2022, ECCV2022, ASE2020, CCS2020, FSE2020

Journal Reviewer: TCSVT, TIP, EMSE

Projects


Responsive image

Personal Opearting System
Implementing the OS from scratch, including BIOS, Driver, Root File System, Segmentation, System Call, Process Support and Shells. You can play the video game"Puzzle" on it. The OS runs on emulator QEMU.

Skills


Related Courses: Learning Theory, Reinforcement Learning, Graph Model, Natural Language Processing, Statistical Machine Learning.
Language: Python, C, C++, SQL, HTML.
Tools: Pytorch, Tensorflow, MXNET.

Contact


Netflix
Los Gatos, CA 95054
xennaughtyboy AT gmail DOT com