Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications.
Boyi WeiKaixuan HuangYangsibo HuangTinghao XieXiangyu QiMengzhou XiaPrateek MittalMengdi WangPeter HendersonPublished in: CoRR (2024)
Keyphrases
- low rank
- missing data
- convex optimization
- linear combination
- matrix completion
- matrix factorization
- singular value decomposition
- rank minimization
- low rank matrix
- high dimensional data
- high order
- matrix decomposition
- kernel matrix
- semi supervised
- trace norm
- singular values
- data matrix
- minimization problems
- computer vision
- robust principal component analysis
- sparse matrix
- multi task
- low dimensional
- higher order
- collaborative filtering
- data analysis
- similarity measure
- feature extraction