1 / 17

Neighbor embedding XOM for dimension reduction and visualization

Neighbor embedding XOM for dimension reduction and visualization. Presenter : Kung, Chien-Hao Authors : Kerstin Bunte , Barbara Hammer, Thomas Villmann , Michael Biehl , Axel Wismuller 2011, NN. Outlines. Motivation Objectives Methodology Experiments Conclusions

durin
Download Presentation

Neighbor embedding XOM for dimension reduction and visualization

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Neighbor embedding XOM for dimension reduction and visualization Presenter : Kung, Chien-HaoAuthors : Kerstin Bunte, Barbara Hammer, Thomas Villmann, Michael Biehl, Axel Wismuller2011, NN

  2. Outlines • Motivation • Objectives • Methodology • Experiments • Conclusions • Comments

  3. Motivation • A novel approach to topology learning: XOM • XOM supports both structure-preserving dimensionality reduction and data clustering. • There is no restriction whatsoever on the distance measures used in XOM.

  4. Objectives • Create a conceptual link between: • Fast sequential online learning known from topology-preserving mappings • Principled direct divergence optimization approaches. • Such as SNE and t-SNE

  5. Methodology-Framework XOM SNE Define as the best match input vector Adaptation rule t-SNE

  6. Methodology Gaussian neighborhood T-distribution

  7. Methodology

  8. Experiments-Parameter setting

  9. Experiments-Complexity

  10. Experiments

  11. Experiments-USPS data set

  12. Experiments-cat cortex and protein data set

  13. Experiments-cat cortex and protein data set

  14. Experiments-cat cortex and protein data set

  15. Experiments

  16. Conclusions • NE-XOM as a competitive trade-off between: • High embedding quality • Low computational expense • NE-XOM allows the user to incorporate prior knowledge and to adapt the level of detail resolution.

  17. Comments • Advantages • This content is expressed clearly . • Applications • Dimension reduction .

More Related