We are organizing the 1st Unlearning and Model Editing (U&Me) workshop at ECCV'24!

Workshop description

The 1st Unlearning and Model Editing (U&Me) workshop will be a half day at ECCV 2024 in Milan on September 29 and will focus on the growing need for new, efficient, and effective techniques for editing trained models, especially large generative models. Such models have practically unlimited functionality in the output they can generate. To provide this functionality, generative models require massive amounts of data and enormous compute costs to train, making it prohibitively expensive to retrain them whenever the need arises: when safety risks are uncovered, when deploying them to compute or storage restricted platforms, or simply due to changing changing requirements. Ensuring that these models are safe and compliant with regulations, in particular, can be difficult due to their broad range of capabilities and a continuously evolving regulatory landscape.

Topics

This workshop will provide a venue for original scientific work presenting novel model editing techniques. We also encourage submissions on model compression methods, which aim to minimize the storage and computational costs of operating trained models with little impact on their performance. Additionally, papers that explore methods for adapting foundation models for efficient fine-tuning or editing are relevant.

Workshop topics: We solicit papers on topics including, but not limited to:

• Unlearning
• Model Stitching and Editing
• Model compression
• Efficient domain adaptation
• Multi-domain/cross-domain U&ME
• Online/lifelong learning, unlearning, and model editing
• Responsible U&ME (e.g., robustness, ethics and fairness, resource efficiency, privacy, and regulatory compliance)
• Applications in computer vision

Though we prioritize papers on generative models, but we also welcome submissions that present methods designed for other types of models, such as discriminative classifiers.

More info at https://sites.google.com/view/u-and-me-workshop/