ALTO: Active Learning with Topic Overviews for Speeding Label Induction and Document Labeling.
Forough Poursabzi-SangdehJordan L. Boyd-GraberLeah FindlaterKevin D. SeppiPublished in: ACL (1) (2016)
Keyphrases
- active learning
- active learning framework
- labeling effort
- label assignment
- version space
- label noise
- document content
- document set
- active learner
- topic discovery
- image labeling
- labeled data
- random sampling
- machine learning
- information retrieval systems
- document images
- active learning strategies
- learning algorithm
- document collections
- information retrieval
- unlabeled instances
- labeling process
- topic hierarchy
- semi supervised
- latent topics
- supervised learning
- document clustering
- relevance feedback
- related documents
- sample selection
- focused crawler
- topic models
- selective sampling
- single document summarization
- learning process
- semi supervised learning
- retrieval systems
- multi label
- document retrieval
- textual content
- data sets
- training set
- unlabeled data
- text documents
- scientific papers
- latent dirichlet allocation
- semantic labels
- relevant documents
- document corpus
- news articles
- document representation
- keywords
- connected component labeling
- transfer learning
- label fusion
- labeling problems
- labeled examples
- multi document summarization
- label propagation