Reading Group: Maninfold Learning & Nonlinear Dimensionality Reduction

SpeakerJonathan Young
AffiliationCMIC, UCL
DateWednesday, 22 Jun 2011
Time14:00 - 15:00
LocationEng 1.13
Event seriesMachine Learning for Neuroimaging Reading Group

PAPER 1: Nonlinear Dimensionality Reduction by Locally Linear Embedding. Abstract: Many areas of science depend on exploratory data analysis and visualization. The need to analyze large amounts of multivariate data raises the fundamental problem of dimensionality reduction: how to discover compact representations of high-dimensional data. Here, we introduce locally linear embedding (LLE), an unsupervised learning algorithm that computes low-dimensional, neighbor- hood-preserving embeddings of high-dimensional inputs. Unlike clustering methods for local dimensionality reduction, LLE maps its inputs into a single global coordinate system of lower dimensionality, and its optimizations do not involve local minima. By exploiting the local symmetries of linear reconstruc- tions, LLE is able to learn the global structure of nonlinear manifolds, such as those generated by images of faces or documents of text. PAPER 2: MANIFOLD LEARNING COMBINING IMAGING WITH NON-IMAGING INFORMATION. Abstract: Recent work suggests that the space of brain magnetic resonance (MR) images can be described by a nonlinear and low-dimensional manifold. In the context of classifying Alzheimer�s disease (AD) patients from healthy controls, we propose a method to incorporate subject meta-information into the manifold learning step. Informa- tion such as gender, age or genotype is often available in clinical studies and can inform the classification of a given query subject. In the proposed method, such information, whether discrete or contin- uous, can be used as an additional input to manifold learning and to enrich a distance measure derived from pairwise image similari- ties. Building on previous work, the Laplacian eigenmap objective function is extended to include the additional information. We use the ApoE genotype, the CSF-concentration of Aβ42 and hippocam- pal volume as meta-information to achieve significantly improved classification results for subjects in the Alzheimer�s Disease Neu- roimaging Initiative (ADNI) database.

iCalendar csml_id_47.ics