Unlabeled data improves word prediction

14 years 9 months ago

Download www.cs.uiuc.edu

Labeling image collections is a tedious task, especially when multiple labels have to be chosen for each image. In this paper we introduce a new framework that extends state of the art models in word prediction to incorporate information from unlabeled examples, using manifold regularization. To the best of our knowledge this is the first semisupervised multi-task model used in vision problems. The new model can be solved using gradient descent and is fast and efficient. We show remarkable improvements for cases with few labeled examples for challenging multi-task learning problems in vision (predicting words for images and attributes for objects).

Nicolas Loeff, Ali Farhadi, Ian Endres and David A

Real-time Traffic

Computer Vision | ICCV 2009 | Word Prediction |

claim paper

» Exploiting Unlabeled Data for Improving Accuracy of Predictive Data Mining

» Boosting Statistical Word Alignment Using Labeled and Unlabeled Data

» SemiSupervised Sequential Labeling and Segmentation Using GigaWord Scale Unlabeled Data

» Semisupervised Prediction of Protein Interaction Sentences Exploiting Semantically Encoded...

» Predicting Subjectivity in Multimodal Conversations

» Efficient Learning by Combining ConfidenceRated Classifiers to Incorporate Unlabeled Medic...

» On Robustness and Domain Adaptation using SVD for Word Sense Disambiguation

» SemiSupervised Conditional Random Fields for Improved Sequence Segmentation and Labeling

Post Info
More Details (n/a)

Added	13 Jul 2009
Updated	10 Jan 2010
Type	Conference
Year	2009
Where	ICCV
Authors	Nicolas Loeff, Ali Farhadi, Ian Endres and David A. Forsyth

Comments (0)

Sciweavers

Unlabeled data improves word prediction

Computer Vision | ICCV 2009 | Word Prediction |

Explore & Download

Productivity Tools

Document Tools

Image Tools

Sciweavers