Scan Statistics over Networks

(6 publications)
We develop novel methodology for detecting faint signals in sensor networks and images. In images we develop a precise asymptotic analysis for fast scanning tools based on convolutional feedforward neural nets. In general graphs, we look at approximation algorithms for the generalized likelihood ratio test.
Exact Asymptotics for the Scan Statistic and Fast Alternatives (arXiv, code on github
J. Sharpnack and E. Arias-Castro
Electronic Journal of Statistics, 2016
Detecting Anomalous Activity on Networks with the Graph Fourier Scan Statistic (arXiv)
J. Sharpnack, A. Rinaldo, and A. Singh
IEEE Transaction on Signal Processing, 2016
Near-optimal Anomaly Detection in Graphs using Lovasz Extended Scan Statistic (arXiv)
J. Sharpnack, A. Krishnamurthy, and A. Singh
Neural Information Processing Systems (NIPS), 2013.
Near-optimal and Computationally Efficient Detectors for Weak and Sparse Graph-structured Patterns (pdf)
J. Sharpnack and A. Singh
IEEE Global Conference on Signal and Information Processing (GlobalSIP), 2013
Detecting Activations over Graphs using Spanning Tree Wavelet Bases (arXiv)
J. Sharpnack, A. Krishnamurthy, and A. Singh
With Oral Presentation
International Conference on Artificial Intelligence and Statistics (AIStats), 2013
Changepoint Detection over Graphs with the Spectral Scan Statistic (arXiv)
J. Sharpnack, A. Rinaldo, and A. Singh
International Conference on Artificial Intelligence and Statistics (AIStats), 2013