TR2008-013

Sparse and Shift-Invariant Feature Extraction From Non-Negative Data

    •  Smaragdis, P., Raj, B., Shashanka, M., "Sparse and Shift-Invariant Feature Extraction from Non-Negative Data", IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), March 2008, pp. 2069-2072.
      BibTeX TR2008-013 PDF
      • @inproceedings{Smaragdis2008mar,
      • author = {Smaragdis, P. and Raj, B. and Shashanka, M.},
      • title = {Sparse and Shift-Invariant Feature Extraction from Non-Negative Data},
      • booktitle = {IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)},
      • year = 2008,
      • pages = {2069--2072},
      • month = mar,
      • issn = {1520-6149},
      • url = {https://www.merl.com/publications/TR2008-013}
      • }
  • Research Areas:

    Artificial Intelligence, Speech & Audio

Abstract:

In this paper we describe a technique that allows the extraction of multiple local shift-invariant features from analysis of non-negative data of arbitrary dimensionality. Our approach employs a probabilistic latent variable model with sparsity constraints. We demonstrate its utility by performing feature extraction in a variety of domains ranging from audio to images and video.

 

  • Related News & Events

    •  NEWS    ICASSP 2008: 4 publications by MERL researchers and others
      Date: March 31, 2008
      Where: IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)
      Brief
      • The papers "Joint Tracking and Video Registration by Factorial Hidden Markov Models" by Mei, X. and Porikli, F., "Speech Denoising Using Nonnegative Matrix Factorization with Priors" by Wilson, K.W., Raj, B., Smaragdis, P. and Divakaran, A., "Ultrasonic Doppler Sensor for Speaker Recognition" by Kalgaonkar, K. and Raj, B. and "Sparse and Shift-Invariant Feature Extraction from Non-Negative Data" by Smaragdis, P., Raj, B. and Shashanka, M. were presented at the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP).
    •