English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Machine unlearning has emerged as a prominent and challenging area ofinterest, driven in large part by the rising regulatory demands for industriesto delete user data upon request and the heightened awareness of privacy.Existing approaches either retrain models from scratch or use severalfinetuning steps for every deletion request, often constrained by computationalresource limitations and restricted access to the original training data. Inthis work, we introduce a novel class unlearning algorithm designed tostrategically eliminate an entire class or a group of classes from the learnedmodel. To that end, our algorithm first estimates the Retain Space and theForget Space, representing the feature or activation spaces for samples fromclasses to be retained and unlearned, respectively. To obtain these spaces, wepropose a novel singular value decomposition-based technique that requireslayer wise collection of network activations from a few forward passes throughthe network. We then compute the shared information between these spaces andremove it from the forget space to isolate class-discriminatory feature spacefor unlearning. Finally, we project the model weights in the orthogonaldirection of the class-discriminatory space to obtain the unlearned model. Wedemonstrate our algorithm's efficacy on ImageNet using a Vision Transformerwith only $\sim$1.5% drop in retain accuracy compared to the original modelwhile maintaining under 1% accuracy on the unlearned class samples. Further,our algorithm consistently performs well when subject to Membership InferenceAttacks showing 7.8% improvement on average across a variety of imageclassification datasets and network architectures, as compared to otherbaselines while being $\sim$6x more computationally efficient.

Deep Unlearning: Fast and Efficient Training-free Approach to Controlled Forgetting