Lacuna Reconstruction: Self-supervised Pre-training for Low-Resource Historical Document Transcription.
Nikolai VoglerJonathan Parkes AllenMatthew Thomas MillerTaylor Berg-KirkpatrickPublished in: CoRR (2021)
Keyphrases
- document retrieval
- information retrieval
- three dimensional
- online learning
- historical data
- resource allocation
- training process
- training samples
- information retrieval systems
- text classifiers
- training examples
- text documents
- keywords
- reconstruction method
- resource constraints
- training phase
- training corpus
- historical documents
- document classification
- document clustering
- image reconstruction
- document images
- document collections
- supervised learning
- active learning
- high resolution
- training set