Metric Learning for Text Documents

High dimensional structured data such as text and images is often poorly understood and misrepresented in statistical modeling. The standard histogram representation suffers from high variance and performs poorly in general. We explore novel connections between statistical translation, heat kernels on manifolds and graphs, and expected distances. These connections provide a new framework for unsupervised metric learning for text documents. Experiments indicate that the resulting distances are generally superior to their more standard counterparts.

• Type: Journal
• Year: 2006
• Where: PAMI
• Authors: Guy Lebanon
• Keywords: Heat Kernels, PAMI 2006, Standard Histogram Representation, Statistical Modeling