Read the original document by opening this link in a new tab.
Table of Contents
1. Abstract 2. Introduction 3. Related Work 4. Model Formulation
Summary
The paper explores the problem of learning to rank from multiple information sources through multi-view learning. It introduces a generic framework for multi-view subspace learning to rank (MvSL2R) and presents novel solutions under this framework. The proposed method is evaluated on university ranking, multi-view lingual text ranking, and image data ranking, showing superior results. The paper discusses the importance of learning to rank in various applications and highlights the need for appropriate data representation and scoring functions. It also delves into the concepts of multi-view learning, subspace learning, and co-training in the context of ranking problems. The proposed methods, including Deep Multi-view Canonical Correlation Analysis (DMvCCA) and Deep Multi-view Modular Discriminant Analysis (DMvMDA), aim to enhance feature discriminability and optimize ranking performance using multiple views. The paper concludes by presenting an end-to-end ranking solution that combines separate view objectives into a joint ranking objective for improved performance.