Matches in ScholarlyData for { ?s ?p This work builds on and responds to previous publications on adaptation of similarity measures to user voting data from the MagnaTagATune dataset. The similarity dataset presented by Stober and Nürnberger at AMR 2011 has been reproduced to test other approaches in a comparable way. On this set, we compare their two-level approach, defining similarity measures on individual facets and combining them in a linear model, to the Metric Learning to Rank (MLR) algorithm which adapts a measure that operates directly on low-level features. We compare the different algorithms, features and parameter spaces with regards to minimising constraint violations. Furthermore, the effectiveness of the MLR algorithm in generalising over unknown similarity data is evaluated on this dataset. We explore the effects of feature choice. Here, we found that the binary genre data showed little correlation with the similarity data, but combined with audio features it clearly improved generalisation.. }
Showing items 1 to 1 of
1
with 100 items per page.
- 9 abstract "This work builds on and responds to previous publications on adaptation of similarity measures to user voting data from the MagnaTagATune dataset. The similarity dataset presented by Stober and Nürnberger at AMR 2011 has been reproduced to test other approaches in a comparable way. On this set, we compare their two-level approach, defining similarity measures on individual facets and combining them in a linear model, to the Metric Learning to Rank (MLR) algorithm which adapts a measure that operates directly on low-level features. We compare the different algorithms, features and parameter spaces with regards to minimising constraint violations. Furthermore, the effectiveness of the MLR algorithm in generalising over unknown similarity data is evaluated on this dataset. We explore the effects of feature choice. Here, we found that the binary genre data showed little correlation with the similarity data, but combined with audio features it clearly improved generalisation.".