WIPE-OUT 2026: Machine Unlearning Workshop at ECML-PKDD – 2nd Workshop on Machine Unlearning and Privacy Preservation
Machine Unlearning Conference Workshop — Co-located with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD 2026), September 7th, 2026, Naples, Italy
Organisers
Andrea D'Angelo
Claudio Savelli
Dr. Flavio Giobergia
Francesco Gullo
Prof. Giovanni Stilo
Table of contents
- Aims and Scope
- Important Dates
- Topics of Interests
- Invited Speakers
- Program at a Glance
- Submission and Publication
- Program Committee
- Registration and Presentation Policy
- Organizers Contacts
About WIPE-OUT 2026
WIPE-OUT 2026 (Workshop on Machine Unlearning and Privacy Preservation) is an academic workshop dedicated to machine unlearning — the process of selectively removing the influence of specific training data from machine learning models without retraining from scratch. The workshop is co-located with ECML-PKDD 2026, one of Europe’s leading machine learning conferences, and will be held on September 7, 2026 in Naples, Italy.
- What
- 2nd Workshop on Machine Unlearning and Privacy Preservation (WIPE-OUT 2)
- When
- September 7, 2026
- Where
- Naples, Italy (co-located with ECML-PKDD 2026)
- Paper deadline
- June 5, 2026
- Proceedings
- Springer CCIS post-proceedings, LNCS format (indexed on Google Scholar, DBLP, Scopus)
- First edition
- WIPE-OUT 2025, held at ECML-PKDD 2025, Porto, Portugal (~50 participants)
Direct New Submission Link: WIPE-OUT@CMT - Deadline June 5, 2026
Aims and Scope
As AI adoption soars, so do concerns over data privacy, ethics, and regulatory compliance (e.g., GDPR, AI Act, CCPA, Bill C-27). Machine unlearning emerges as a critical capability, enabling selective removal of learned information without costly retraining — mitigating biases, protecting sensitive data, and aligning AI with ethical standards. Unlike differential privacy, which limits what a model learns during training, machine unlearning addresses the post-hoc removal of data influence, making it essential for enforcing the right to be forgotten in deployed AI systems.
Building on the success of the first edition at ECML-PKDD 2025 in Porto (~50 participants, keynotes by Google DeepMind researchers), this second edition explores cutting-edge methodologies, theoretical advancements, and practical applications of machine unlearning and related topics, including privacy preservation, that align with evolving regulatory standards and societal expectations. We especially encourage submissions from researchers working on unlearning theory, LLMs and foundation models, computer vision, recommender systems, privacy and compliance, and scalable systems — connecting these communities around the shared challenge of selective data removal.
What makes WIPE-OUT different: unlike traditional workshops, WIPE-OUT features pitch-style paper presentations, head-to-head paper comparison debates, and a closing roundtable bridging academia and industry — designed to maximize interaction and constructive discussion.
Important Dates
- Paper Submissions: June 5, 2026
- Notifications: July 5th, 2026
- Camera-Ready: July 10th, 2026
- Workshop: September 7th, 2026
All deadlines are 11:59 pm AoE.
Machine Unlearning Topics of Interest
This machine unlearning workshop welcomes contributions on all topics related to machine unlearning across domains (e.g., finance, business, basic sciences, construction, computational advertising, medical, etc.) and independent of data types (e.g., networks, tabular, unstructured, graphs, logs, spatiotemporal, multimedia, time series, genomic sequences, and streaming data). Contributions can also include research or perspectives regarding the following:
- Foundations and Theory of Machine Unlearning
- Theoretical foundations, guarantees, and bounds for Machine Unlearning
- Certified and approximate unlearning: complexity, information-theoretic limits, and impossibility results
- Development of new Machine Unlearning and Model Editing Algorithms
- Connections between unlearning, influence functions, and data attribution
- Unlearning in LLMs and Foundation Models
- Knowledge removal and model editing in large language models
- Unlearning unsafe, copyrighted, or private content from pretrained LLMs
- Concept erasure and safety alignment via unlearning in generative models
- Unlearning in multimodal foundation models (vision-language, text-to-image)
- Unlearning in Computer Vision
- Class-level and instance-level forgetting in image classifiers and object detectors
- Concept erasure in diffusion models and GANs
- Unlearning for face recognition, de-identification, and visual privacy
- Continual and incremental unlearning in vision systems
- Unlearning in Recommender Systems
- User data removal in collaborative filtering and content-based recommenders
- Unlearning in sequential, session-based, and graph-based recommendation
- Impact of unlearning on recommendation quality and fairness
- Right-to-be-forgotten compliance in deployed recommendation platforms
- Privacy and Compliance
- Privacy-preserving techniques for Machine Learning systems
- Interactions between differential privacy, federated learning, and unlearning
- Federated and decentralized unlearning processes
- Membership inference, data auditing, and privacy verification after unlearning
- Regulatory compliance (GDPR, AI Act, CCPA) and the right to be forgotten
- Systems and Scalability
- Efficient and scalable machine unlearning in big data systems
- Real-time and streaming data unlearning systems
- Systems-level design for unlearning-ready ML pipelines
- Hardware-aware and resource-efficient unlearning methods
- Evaluation and Benchmarking
- Evaluation metrics for quantifying unlearning performance
- Development of new evaluation protocols for Machine Unlearning
- Tools and benchmarks for unlearning frameworks
- Robustness and verification of unlearning and editing guarantees
- Implications of Machine Unlearning
- Unlearning for Explainable and Interpretable AI
- Ethical, legal and societal aspects of Machine Unlearning
- Machine Unlearning in High-Stakes Applications (e.g., healthcare, finance, …)
Invited Speakers
We plan to have up to two keynote speakers from leading institutions in Machine Unlearning and Privacy Preservation. The invited speakers will be announced shortly — stay tuned!
Sijia Liu - TBA
TBA
|
Speaker's Bio:Sijia Liu is a Red Cedar Distinguished Associate Professor in the Department of Computer Science and Engineering at Michigan State University (MSU), and an Affiliated Professor at the MIT-IBM Watson AI Lab, IBM Research. His research centers on scalable and trustworthy AI, such as machine unlearning for vision and language models, scalable optimization for deep models, adversarial robustness, and data–model efficiency. He has received numerous honors, including the NSF CAREER Award (2024), INNS Aharon Katzir Young Investigator Award (2024), MSU Withrow Rising Scholar Award (2025), Best Paper Runner-Up at UAI (2022), and Best Student Paper Award at ICASSP (2017). He also co-founded the New Frontiers in Adversarial Machine Learning Workshop series (ICML/NeurIPS 2021–2024). |
Preliminary Program at a Glance
WIPE-OUT 2026 will be a half-day event combining keynote presentations, thematic paper sessions with pitch-style talks, interactive “head-to-head” paper comparison debates, poster sessions, and a closing roundtable with experts from academia and industry.
- 14:00 — Welcome — 10 min
- 14:10 — First Thematic Session — 80 min
- Keynote Speaker (1st) — 40 min
- Pitches of best papers for this theme — 40 min
- 15:30 — Posters, Coffee & Break — 30 min
- 16:00 — Second Thematic Session — 80 min
- Keynote Speaker (2nd) — 40 min
- Pitches of best papers for this theme — 40 min
- 17:20 — Comparison Sessions — 20 min
- Papers Comparison on 1st Topic — 10 min
- Papers Comparison on 2nd Topic — 10 min
- 17:40 — Round Table Discussion (Academia & Industry) — 30 min
- 18:10 — Best Paper Awards — 10 min
- 18:20 — Concluding Remarks — 10 min
Submission and Publication
Submission system: WIPE-OUT @ CMT
From the submission system, select the “WIPE-OUT 2 Workshop on Machine Unlearning and Privacy Preservation” track to create a new submission.
Why Submit to WIPE-OUT 2026?
- Post-proceedings published in Springer CCIS (LNCS format), indexed on Google Scholar, DBLP, and Scopus
- Selected papers may be invited to a special journal issue
- Best Paper Awards for outstanding contributions
- Interaction with leading researchers from institutions such as Google DeepMind, University of Toronto, NUS, and more
- A unique format with pitch talks, head-to-head debates, and an academia-industry roundtable
We invite authors to submit unpublished, original papers written in English. Submitted papers should not have been previously published or accepted for publication in substantially similar form in any peer-reviewed venue, such as journals, conferences, or workshops. Papers should be prepared and submitted in the LNCS format using the Springer LNCS template.
We will consider three different submission types:
- Full papers (up to 14 pages) should clearly describe the state of the art and state the proposal's contribution in the application domain, even if presenting preliminary results. In particular, research papers should describe the methodology in detail, experiments should be repeatable, and a comparison with the existing approaches in the literature should be made.
- Reproducibility papers (up to 14 pages) should repeat prior experiments using the original source code and datasets to show how, why, and when the methods work or not (replicability papers) or should repeat prior experiments, preferably using the original source code in new contexts (e.g., different domains and datasets, different evaluation and metrics) to generalize further and validate or not previous work (reproducibility papers).
- Short, Demo, and position papers (up to 8 pages) should introduce new points of view on the workshop topics or summarize a group's experience in the field. Practice and experience reports must detail real-world scenarios in which Machine Unlearning is needed.
Unlimited appendix are accepted at submission time.
Submissions should not exceed the indicated pages, including any diagrams and references. All submissions will undergo a double-blind review process and be reviewed by at least three reviewers based on relevance to the workshop, novelty/originality, significance, technical quality and correctness, quality and clarity of presentation, quality of references, and reproducibility. Submitted papers will be rejected without review if they are not correctly anonymized, do not comply with the template, or do not follow the above guidelines.
Generative AI Usage Policy. Generative AI models, including Chat-GPT, BARD, LLaMA, or similar LLMs, do not satisfy the criteria for authorship of papers accepted in the workshop. If authors use an LLM in any part of the paper-writing process, they assume full responsibility for all content, including checking for plagiarism and correctness of all text.
Proceedings.
The accepted papers and the material generated during the meeting will be available on the workshop website. The workshop proceedings will be published as a joint post-workshop proceeding in Springer’s Communications in Computer and Information Science (CCIS), in 1-2 volumes organized by focused scope, indexed on Google Scholar, DBLP, and Scopus. Paper authors will have the faculty to opt-in or opt-out. The authors of selected papers may be invited to submit an extended version to a special journal issue.
Program Committee
Click to expand the full list of program committee members
- Hongzhi Yin
- Ximing Lu
- Wei Yuan
- Jiaao Chen
- Jiancheng Liu
- S. Welleck
- Xiaojun Xu
- Diyi Yang
- Sijia Liu
- Liwei Jiang
- Shashwat Goel
- Pratyush Maini
- Jinghan Jia
- Jack Hessel
- P. Kumaraguru
- Zhili Feng
- Mohan S. Kankanhalli
- Lianhui Qin
- Ronen Eldan
- Avi Schwarzschild
- Vikram S Chundawat
- Peter West
- M. Russinovich
- Zachary Chase Lipton
- Ayush K Tarun
- Prithviraj Ammanabrolu
- O. Milenkovic
- J. Kolter
- Gaowen Liu
- Yejin Choi
- Eli Chien
- Parikshit Ram
- Murari Mandal
- Phi-Le Nguyen
- Yi Liu
- Chen Wu
- Tomasz Korbak
- Alan Wee-Chung Liew
- Lei Xu
- Sencun Zhu
- Kejian Shi
- Yuguang Yao
- Xingliang Yuan
- P. Mitra
- Angelica Chen
- Sungmin Cha
- Cong Wang
- Ambrish Rawat
- Rasika Bhalerao
- Moontae Lee
- Yimeng Zhang
- S. Kadhe
- C. L. Buckley
- Nicolas Papernot
- Yiwen Jiang
- Alice Gatti
- Jason Phang
- Joel Jang
- Shenglong Liu
- Rishub Tamirisa
- Sam Bowman
- Dongkeun Yoon
- Tao Zhao
- Bhrugu Bharathi
- Ethan Perez
- Sohee Yang
- Wei Li
- Andy Zou
- Q. Nguyen
- Lajanugen Logeswaran
- Xianzhou Gao
- Mantas Mazeika
- Yihua Zhang
- Minjoon Seo
- Matthew Jagielski
- P. Triantafillou
- T. Huynh
- Eleni Triantafillou
- Chongyu Fan
- Nathalie Baracaldo
- T. Nguyen
- Yuanshun Yao
- M. Kurmanji
- Stephen Casper
Registration and Presentation Policy
Please be aware that at least one author of all accepted papers should register to the conference until the early registration deadline.
We expect the authors, the program committee, and the organizing committee to adhere to the ECML-PKDD Code of Conduct.
The Main Conference organization team will manage the registration: https://ecmlpkdd.org/2026/
Frequently Asked Questions
What is machine unlearning?
Machine unlearning is the process of selectively removing the influence of specific training data points from a trained machine learning model, without retraining the model from scratch. It addresses the need to comply with data deletion regulations such as the GDPR's "right to be forgotten," to correct biases, and to remove sensitive or outdated information from deployed AI systems.
Why attend a machine unlearning workshop?
A dedicated machine unlearning workshop like WIPE-OUT brings together researchers and practitioners focused specifically on data removal, privacy preservation, and model editing. Unlike broader machine learning conferences, the workshop offers focused discussions, peer review from domain experts, and networking with the growing machine unlearning community.
Is WIPE-OUT 2026 a machine unlearning conference or a workshop?
WIPE-OUT 2026 is a workshop co-located with the ECML-PKDD 2026 conference. While it is not a standalone conference, it functions as a dedicated machine unlearning venue within one of Europe's top machine learning conferences. Accepted papers are published in Springer CCIS post-proceedings (LNCS format) and indexed on Google Scholar, DBLP, and Scopus.
How does machine unlearning differ from differential privacy?
Differential privacy limits what a model can learn about individual data points during training by adding noise. Machine unlearning, by contrast, operates after training — it removes or reduces the influence of specific data that the model has already learned. The two techniques are complementary: differential privacy provides proactive protection, while machine unlearning enables reactive data removal.
What topics are covered at WIPE-OUT 2026?
The workshop covers theoretical foundations of machine unlearning, unlearning in large language models and foundation models, concept erasure in computer vision and generative models, unlearning in recommender systems, privacy-preserving techniques and regulatory compliance, scalable systems for efficient unlearning, evaluation metrics and benchmarks, and the ethical, legal, and societal implications of machine unlearning.
Organizers Contacts
For general inquiries about the machine unlearning workshop, please email andrea.dangelo6@graduate.univaq.it, claudio.savelli@polito.it, flavio.giobergia@polito.it, gullof@acm.org, and gstilo@luiss.it.