Noise-Robust Dense Retrieval via Contrastive Alignment Post Training.
Daniel CamposChengXiang ZhaiAlessandro MagnaniPublished in: CoRR (2023)
Keyphrases
- training set
- image noise
- supervised learning
- active learning
- image database
- information retrieval
- image retrieval
- greater robustness
- training data
- salt pepper
- noise model
- arbitrary shape
- noise level
- noisy environments
- random noise
- information retrieval systems
- content based retrieval
- retrieval accuracy
- noisy data
- document retrieval
- geometric distortions
- imaging artifacts
- relevance feedback