Foundations and Theoretical Guarantees
Foundations, formulations and theoretical guarantees of machine unlearning
ICVGIP 2026 · Workshop
Call for Papers
MULVA invites researchers and practitioners to explore how AI systems can forget specified information while retaining the knowledge, capabilities and reliability that matter
Submission Information
Submission deadlines, paper formats, publication arrangements and submission links will be announced in accordance with ICVGIP 2026 policies. Please revisit this page for the latest confirmed information
Submission details will be updated when confirmedExplore the research landscape
Topics of Interest
We welcome research spanning the theory, algorithms, evaluation and practical deployment of machine unlearning across language, vision and multimodal AI
Unlearning Methods and Algorithms
Exact, approximate, certified and parameter-efficient unlearning methods
Granularity of Forgetting
Data-, instance-, class-, concept-, identity- and knowledge-level unlearning
Vision and Language Models
Unlearning in computer vision, NLP, LLMs and visual foundation models
Multimodal and Generative AI
Unlearning in multimodal, vision-language, generative and foundation models
Model Editing and Knowledge Removal
Model editing, knowledge removal and their relationship to machine unlearning
Scalable and Continual Unlearning
Efficient, scalable, federated, distributed and continual unlearning
Privacy and Responsible AI
Privacy, copyright, data deletion, safety, fairness and responsible AI
Security and Verification
Adversarial attacks, membership inference, privacy leakage and verification of forgetting
Evaluation and Benchmarking
Evaluation metrics, benchmarks, datasets and forgetting–retention trade-offs
Multimodal Agents and AI Systems
Unlearning for multimodal agents and real-world AI systems
Deployment and Open Challenges
Industrial deployment, lifecycle management and open challenges in trustworthy computational unlearning
Share your research
Contribution Types
Contributions may address machine unlearning from theoretical, empirical, application-oriented or critical perspectives
Algorithms and Theory
Novel unlearning algorithms, formal formulations and theoretical analyses
Empirical Evaluations
Empirical studies, systematic comparisons and experimental evaluations
Vision, Language and Multimodal AI
Unlearning studies involving computer vision, natural language processing and multimodal models
Datasets and Benchmarks
Datasets, benchmark suites and evaluation protocols for measuring unlearning
Privacy and Security
Analyses of privacy, security, attacks, information leakage and forgetting verification
Position and Perspective
Critical perspectives, open research questions and future directions for machine unlearning
MULVA 2026 · ICVGIP
What should AI learn to forget?
Join the research community working towards reliable, efficient and verifiable machine unlearning