Abstract

A Spatio-temporal Extension to Isomap Nonlinear Dimension Reduction
Odest Jenkins - University of Southern California Maja Mataric - University of Southern California
We present an extension of Isomap nonlinear dimension reduction (Tenenbaum etal. 2000) for data with both spatial and temporal relationships. Our method,ST-Isomap, augments the existing Isomap framework to consider temporalrelationships in local neighborhoods that can be propagated globally via ashortest-path mechanism. Two instantiations of ST-Isomap are presented forsequentially continuous and segmented data. Results from applying ST-Isomapto real-world data collected from human motion performance and humanoid robotteleoperation are also presented.

A Spatio-temporal Extension to Isomap Nonlinear Dimension Reduction

Odest Jenkins - University of Southern California
Maja Mataric - University of Southern California

We present an extension of Isomap nonlinear dimension reduction (Tenenbaum etal. 2000) for data with both spatial and temporal relationships. Our method,ST-Isomap, augments the existing Isomap framework to consider temporalrelationships in local neighborhoods that can be propagated globally via ashortest-path mechanism. Two instantiations of ST-Isomap are presented forsequentially continuous and segmented data. Results from applying ST-Isomapto real-world data collected from human motion performance and humanoid robotteleoperation are also presented.