A Sliding Information Distance for Change Point Detection in Text or Audio

• Activity Number: 301 - Natural Language Processing Applications in Defense and National Security
• Type: Topic Contributed
• Date/Time: Wednesday, August 5, 2020 : 10:00 AM to 11:50 AM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Richard Field* and Christina Ting and Travis Bauer
• Companies: Sandia National Laboratories and Sandia National Laboratories and Sandia National Laboratories
• Keywords: Information theory; Information distance; Change detection

Abstract

We propose a general method for change point detection within text or audio signals. Our approach is based on an approximation of the universal normalized information distance, referred to as the sliding information distance (SLID). When combined with a wavelet-based approach for peak identification, SLID provides an efficient, accurate, and robust method for change point detection. We apply SLID to several datasets to demonstrate the efficacy of the approach.

Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

SAND2020-1404 A