150 likes | 207 Views
Modeling Semantic Similarities in Multiple Maps. Presenter : Wei- Hao Huang Authors : Laurens van der Maaten , Geoffrey Hinton EWI-ICT TR, 2009. Outlines. Motivation Objectives Methodology Experiments Conclusions Comments. Motivation.
E N D
Modeling Semantic Similarities inMultiple Maps Presenter : Wei-Hao Huang Authors : Laurens van der Maaten, Geoffrey Hinton EWI-ICT TR, 2009
Outlines • Motivation • Objectives • Methodology • Experiments • Conclusions • Comments
Motivation • Semantic space models cannot faithfully represent intransitive pairwise similarities or the similarities of words that have multiple meanings. • Triangle inequality • Nearest neighbor is limited • Similarities are symmetric suit dog dog tuxedo Animal tie dog rope knot dog dog North Korea China
Objectives • To propose multiple map SNE to solve fundamental limitations of metric spaces suit tuxedo tie suit tuxedo tie rope knot rope tie knot
Methodology Data Data Mixing proportion (importance) SNE Multiple maps SNE Map3 Map2 Map Map1 Stochastic neighbor embedding Multiple maps SNE
Methodology suit tuxedo tie rope knot Stochastic neighbor embedding
Methodology dog dog dog dog suit rope animal animal tie tie dog dog Multiple map SNE
Methodology A*C=1*1/2=1/2 B*C=1*1/2=1/2 Multiple map SNE
Experiments • Visualization Experiments • Florida State University word association dataset • Selecting 5019 words • Generalization Experiments • To evidence their model for semantic representation • Training data: 80% • Validation data: 10% • Test data: 10%
Experiments Visualization
Experiments Statue of Liberty Sport Cheerleader Clothing monarchy Tie Tie Cheerleader monarchy
Experiments Generalization
Experiments • Comparing multiple maps SNE with other method. • Semantic space models • Semantic networks • Topic models
Conclusions The multiple maps SNE alleviates the fundamental limitations of metric spaces. Multiple map model has characteristics that are similar to those of topic models.
Comments • Advantages • Multiple maps SNE alleviates the fundamental limitations of metric spaces • Applications • Data visualization • Semantic similarities