Improving peak picking using multiple time-step loss functions

Southall, Carl and Stables, Ryan and Hockman, Jason (2018) Improving peak picking using multiple time-step loss functions. In: International Society of Music Information Retrieval Conference, 23rd-27th September, 2018, Paris, France.

[img]
Preview
Text
CSRSJH_LossFunctions_Final (1).pdf - Published Version
Available under License Creative Commons Attribution.

Download (4MB)

Abstract

The majority of state-of-the-art methods for music infor-mation retrieval (MIR) tasks now utilise deep learningmethods reliant on minimisation of loss functions such ascross entropy. For tasks that include framewise binaryclassification (e.g., onset detection, music transcription)classes are derived from output activation functions byidentifying points of local maxima, or peaks. However, theoperating principles behind peak picking are different tothat of the cross entropy loss function, which minimises theabsolute difference between the output and target valuesfor a single frame. To generate activation functions moresuited to peak-picking, we propose two versions of a newloss function that incorporates information from multipletime-steps: 1)multi-individual, which uses multiple indi-vidual time-step cross entropies; and 2)multi-difference,which directly compares the difference between sequentialtime-step outputs. We evaluate the newly proposed lossfunctions alongside standard cross entropy in the popularMIR tasks of onset detection and automatic drum tran-scription. The results highlight the effectiveness of theseloss functions in the improvement of overall system ac-curacies for both MIR tasks. Additionally, directly com-paring the output from sequential time-steps in the multi-difference approach achieves the highest performance.

Item Type: Conference or Workshop Item (Paper)
Dates:
DateEvent
25 May 2018Accepted
27 September 2018Published Online
Subjects: CAH11 - computing > CAH11-01 - computing > CAH11-01-04 - software engineering
CAH11 - computing > CAH11-01 - computing > CAH11-01-05 - artificial intelligence
CAH25 - design, and creative and performing arts > CAH25-02 - performing arts > CAH25-02-02 - music
Divisions: Faculty of Computing, Engineering and the Built Environment > College of Computing
Depositing User: Jason Hockman
Date Deposited: 13 Aug 2018 08:22
Last Modified: 19 Jun 2024 12:39
URI: https://www.open-access.bcu.ac.uk/id/eprint/6186

Actions (login required)

View Item View Item

Research

In this section...