English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Inspired by the success of deploying deep learning in the fields of ComputerVision and Natural Language Processing, this learning paradigm has also foundits way into the field of Music Information Retrieval. In order to benefit fromdeep learning in an effective, but also efficient manner, deep transferlearning has become a common approach. In this approach, it is possible toreuse the output of a pre-trained neural network as the basis for a newlearning task. The underlying hypothesis is that if the initial and newlearning tasks show commonalities and are applied to the same type of inputdata (e.g. music audio), the generated deep representation of the data is alsoinformative for the new task. Since, however, most of the networks used togenerate deep representations are trained using a single initial learningsource, their representation is unlikely to be informative for all possiblefuture tasks. In this paper, we present the results of our investigation ofwhat are the most important factors to generate deep representations for thedata and learning tasks in the music domain. We conducted this investigationvia an extensive empirical study that involves multiple learning sources, aswell as multiple deep learning architectures with varying levels of informationsharing between sources, in order to learn music representations. We thenvalidate these representations considering multiple target datasets forevaluation. The results of our experiments yield several insights on how toapproach the design of methods for learning widely deployable deep datarepresentations in the music domain.

One Deep Music Representation to Rule Them All? : A comparative analysis of different representation learning strategies