论文信息 - Large-Scale Graph-based Transductive Inference

Large-Scale Graph-based Transductive Inference

We consider the issue of scalability of graph-based semi-supervised learning (SSL) algorithms. In this context, we propose a fast graph node ordering algorithm that improves (parallel) spatial locality by being cache cognizant. This approach allows for a near linear speedup on a shared-memory parallel machine to be achievable, and thus means that graph-based SSL can scale to very large data sets. We use the above algorithm in a multi-threaded multi-core implementation to solve a SSL problem on a 120 million node graph in a reasonable amount of time.

Amarnag Subramanya | A. Subramanya

[1] Ivor W. Tsang,et al. Large-Scale Sparsified Manifold Regularization , 2006, NIPS.

[2] Zoubin Ghahramani,et al. Combining active learning and semi-supervised learning using Gaussian fields and harmonic functions , 2003, ICML 2003.

[3] Jeff A. Bilmes,et al. Entropic Graph Regularization in Non-Parametric Semi-Supervised Classification , 2009, NIPS.

[4] Nicolas Le Roux,et al. Efficient Non-Parametric Function Induction in Semi-Supervised Learning , 2004, AISTATS.

[5] Xiaojin Zhu,et al. --1 CONTENTS , 2006 .

[6] Jason Weston,et al. Large scale manifold transduction , 2008, ICML '08.

[7] Aren Jansen,et al. Semi-supervised learning of speech sounds , 2007, INTERSPEECH.

[8] Jeff A. Bilmes,et al. The semi-supervised switchboard transcription project , 2009, INTERSPEECH.

[9] Nurul I Sarkar,et al. Keynote speech I , 2011, TENCON 2011 - 2011 IEEE Region 10 Conference.