Multi-Target Bayes Filtering with Random Finite Sets

Add to your list(s) Download to your calendar using vCal

• Paul Peeling
• Friday 15 December 2006, 15:30-17:00
• Engineering Department, Baker Building, Division F meeting room, 5th floor.

If you have a question about this talk, please contact Taylan Cemgil.

Random finite sets are a natural framework for tracking an unknown number of moving targets. Explicit association between measurements and targets can be avoided by adopting a point-process formalisim. This leads to a Bayes multi-target filter with a SMC implementation. Practically this is inefficient for many targets. The Probability Hypothesis Density (PHD) approximates the multi-target posterior in the single-target space, and is more tractable to compute in an SMC implementation.

I will cover the above theory and implementation, and motivate its application to automatic polyphonic music transcription

Vo B-N., Singh S.S., and Doucet A., Sequential Monte Carlo methods for Multi-target Filtering with Random Finite Sets, IEEE Aerospace and Electronic Systems, June 2005 http://www-sigproc.eng.cam.ac.uk/%7Esss40/papers/Vo05_smcForMultiTargetFilteringWthRandomSets.pdf

This talk is part of the Audio and Music Processing (AMP) Reading Group series.

This talk is included in these lists:

• All Talks (aka the CURE list)
• Audio and Music Processing (AMP) Reading Group
• Cambridge University Engineering Department Talks
• Cambridge talks
• Centre for Smart Infrastructure & Construction
• Computational Continuum Mechanics Group Seminars
• Engineering Department, Baker Building, Division F meeting room, 5th floor
• Featured lists
• Interested Talks
• School of Technology
• Trust & Technology Initiative - interesting events
• bld31

Note that ex-directory lists are not shown.