MLSP Prizes for Reproducibility: Winners announced!

With thanks to the IEEE Signal Processing Society and to all of the entrants, we are now able to announce the winners of the MLSP 2014 and SoundSoftware.ac.uk Prizes for Reproducibility in Signal Processing.

As with our earlier prize series in conjuction with the Audio Engineering Society's 53rd Conference and our SoundSoftware.ac.uk Prizes for Reproducibility in Audio and Music Research, you can refer to our page about evaluation for more details about how we came to a decision.

Note that all of the submissions we received were in the Fully-Reproducible Works category, so we did not award a prize for Reproducibility-Enabling Work.

Winner

The winner of the MLSP 2014 and SoundSoftware.ac.uk Prize for Reproducibility in Signal Processing is:

Simo Särkkä and Robert Piché
On Convergence And Accuracy Of State-Space Approximations Of Squared Exponential Covariance Functions (paper, software)

This submission scored highly across all evaluation criteria. The software is clearly presented, with an appropriate licence and a sufficiently informative README and metadata, and hosted on a widely recognised hosting platform. The relationship between the software and paper publication is clear. The code (in MATLAB) is straightforwardly organised and was found to work as described and to replicate the figures presented in the paper.

Honourable mention

An honourable mention goes to:

Aleksandr Aravkin, Karthikeyan Natesan Ramamurthy, and Gianluigi Pillonetto
Kalman Smoothing With Persistent Nuisance Parameters (software)

This is a very creditable submission. The software is well organised and is presented with an appropriate licence through a widely recognised hosting platform. The code (in MATLAB) was found to work as describe, although some paths needed to be adjusted for the testing environment, and to replicate the figures presented in the paper. We would suggest that the README file could be expanded and the relationship between code and paper publication(s) made more explicit, but this is an impressive work nonetheless.