Embedding contrastive unsupervised features to cluster in- and out-of-distribution noise in corrupted image datasets.
Paul AlbertEric ArazoNoel E. O'ConnorKevin McGuinnessPublished in: CoRR (2022)
Keyphrases
- feature vectors
- spatial distribution
- feature set
- feature extraction
- unsupervised feature selection
- noise free
- noise level
- spatial information
- image restoration
- low level
- feature space
- clustering algorithm
- co occurrence
- missing data
- supervised learning
- noise reduction
- image features
- additive gaussian noise
- poisson noise