Graph Structured Signal Processing

(6 publications)
Traditional signal processing techniques such as wavelet denoising and kernel smoothing, implicitly assume that the domain is homogeneous. Graph structure provides us with a more flexible framework for filters over heterogeneous media, social networks, or semantic information.
The DFS fused lasso: nearly optimal linear-time denoising over graphs and trees (arXiv)
Oscar Hernan Madrid Padilla, James G. Scott, James Sharpnack, Ryan J. Tibshirani
In Submission, 2016
Trend Filtering on Graphs (arXiv)
Y. Wang, J. Sharpnack, A. Smola, and R. J. Tibshirani
Journal of Machine Learning Research (JMLR), 2016 and
International Conference on Artificial Intelligence and Statistics (AIStats), 2015
A Path Algorithm for Localizing Anomalous Activity in Graphs (pdf)
J. Sharpnack
IEEE Global Conference on Signal and Information Processing (GlobalSIP), 2013
Recovering Graph-Structured Activations using Adaptive Compressive Measurements (arXiv)
A. Krishnamurthy, J. Sharpnack, and A. Singh
Best Student Paper Award
Asilomar Conference on Signals, Systems, and Computers, 2013
Sparsistency of the Edge Lasso over Graphs (pdf)
J. Sharpnack, A. Rinaldo, and A. Singh
International Conference on Artificial Intelligence and Statistics (AIStats), 2012.
Identifying Graph-structured Activation Patterns in Networks (pdf)
J. Sharpnack, and A. Singh
With Oral Presentation
Neural Information Processing Systems (NIPS), 2010