Chasing the Metric: Smoothing Learning Algorithms for Keyword Detection
Title | Chasing the Metric: Smoothing Learning Algorithms for Keyword Detection |
Publication Type | Conference Paper |
Year of Publication | 2014 |
Authors | Vinyals, O., & Wegmann S. |
Other Numbers | 3663 |
Abstract | In this paper we propose to directly optimize a discrete objectivefunction by smoothing it, showing it is both effective at enhancingthe figure of merit that we are interested in while keeping the over-all complexity of the training procedure unaltered. We looked at thetask of keyword detection with data scarcity (e.g., for languages forwhich we do not have enough data), and found it useful to optimizethe Actual Term Weighted Value (ATWV) directly. In particular, we |
Acknowledgment | This work was supported by the Intelligence Advanced Research Projects Activity (IARPA) via Department of Defense US Army Research Laboratory contract number W911NF-12-C-0014 ("Spoken WOrdsearch with Rapid Development and Frugal Invariant Subword Hierarchies - Swordfish"). The U.S. Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright annotation thereon. Disclaimer: The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of IARPA, DoD/ARL, or the U.S. Government. |
URL | https://www.icsi.berkeley.edu/pubs/speech/chasingmetric14.pdf |
Bibliographic Notes | Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2014), Florence, Italy |
Abbreviated Authors | O. Vinyals and S. Wegmann |
ICSI Research Group | Speech |
ICSI Publication Type | Article in conference proceedings |