On Competitiveness of Nearest-Neighbor-Based Music Classification: A Methodological Critique

Pálmason, Haukur; Jónsson, Björn Þór; Amsaleg, Laurent; Schedl, Markus; Knees, Peter

doi:10.1007/978-3-319-68474-1_19

Cited by 6 publications

(5 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Consequently, datasets and data sources normally employed in MGC research do not share a common structure, presenting different classifications and a variable number of genre labels. Moreover, there is no commonly agreed standard dataset for the matter [44], and many of the published research works use private datasets. This circumstance does not enable reproducible results [42].…”

Section: State Of the Artmentioning

confidence: 99%

See 1 more Smart Citation

Web-Based Music Genre Classification for Timeline Song Visualization and Analysis

Castillo¹,

Flores²

2021

IEEE Access

View full text Add to dashboard Cite

This paper presents a web application that retrieves songs from YouTube and classifies them into music genres. The tool explained in this study is based on models trained using the musical collection data from Audioset. For this purpose, we have used classifiers from distinct Machine Learning paradigms: Probabilistic Graphical Models (Naive Bayes), Feed-forward and Recurrent Neural Networks and Support Vector Machines (SVMs). All these models were trained in a multi-label classification scenario. Because genres may vary along a song's timeline, we perform classification in chunks of ten seconds. This capability is enabled by Audioset, which offers 10-second samples. The visualization output presents this temporal information in real time, synced with the music video being played, presenting classification results in stacked area charts, where scores for the top-10 labels obtained per chunk are shown. We briefly explain the theoretical and scientific basis of the problem and the proposed classifiers. Subsequently, we show how the application works in practice, using three distinct songs as cases of study, which are then analyzed and compared with online categorizations to discuss models performance and music genre classification challenges.

show abstract

Section: State Of the Artmentioning

confidence: 99%

“…Other popular datasets are the Million Song Dataset (MSD) [48], FMA [49] or MuMu [50]. As new datasets appear, each with its own structure and features, we observe more difficulties in reaching an agreement on genre descriptors [44].…”

Section: State Of the Artmentioning

confidence: 99%

Web-Based Music Genre Classification for Timeline Song Visualization and Analysis

Castillo¹,

Flores²

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…In particular, k-NN is a relatively popular model used for MGC. Pálmason et al (2017a) report the best classification score for a k-NN model obtained in the GTZAN dataset, achieving accuracy scores in the range 80-81%. Iloga et al (2018) also employ k-NNs for MGC, extracting sequential patterns from music and generating music genre taxonomies.…”

Section: Content-based Classificationmentioning

confidence: 99%

“…As a consequence, datasets and data sources used in MGC research do not share a common structure, presenting different classes and a variable number or them. Moreover, there is not a commonly agreed standard dataset for the matter (Pálmason et al, 2017a) and most of the published research use private datasets that do not allow to reproduce results (Sturm, 2012b). Table 1 lists datasets commonly used in MGC research, plus some with the potential to be used in future work.…”

Section: Data Sets and Sourcesmentioning

confidence: 99%

Machine learning for music genre: multifaceted review and experimentation with audioset

Ramírez

Flores

2019

J Intell Inf Syst

View full text Add to dashboard Cite

Music genre classification is one of the sub-disciplines of music information retrieval (MIR) with growing popularity among researchers, mainly due to the already open challenges. Although research has been prolific in terms of number of published works, the topic still suffers from a problem in its foundations: there is no clear and formal definition of what genre is. Music categorizations are vague and unclear, suffering from human subjectivity and lack of agreement. In its first part, this paper offers a survey trying to cover the many different aspects of the matter. Its main goal is give the reader an overview of the history and the current state-of-the-art, exploring techniques and datasets used to the date, as well as identifying current challenges, such as this ambiguity of genre definitions or the introduction of human-centric approaches. The paper pays special attention to new trends in machine learning applied to the music annotation problem. Finally, we also include a music genre classification experiment that compares different machine learning models using Audioset.

show abstract

“…Since their inception, nearest-neighbor methods for classification and regression in metric spaces continue to attract the interest of theoreticians and practitioners alike. On the applied front, this seemingly naive approach remains competitive against more sophisticated methods [9,59,44,12]. On the theoretical front, the most commonly investigated questions involve Bayes consistency and rates of convergence [29,40,21,11].…”

mentioning

confidence: 99%

Universal Bayes consistency in metric spaces

Hanneke¹,

Kontorovich²,

Sabato³

et al. 2019

Preprint

View full text Add to dashboard Cite

We show that a recently proposed 1-nearest-neighbor-based multiclass learning algorithm is universally strongly Bayes consistent in all metric spaces where such Bayes consistency is possible, making it an "optimistically universal" Bayes-consistent learner. This is the first learning algorithm known to enjoy this property; by comparison, k-NN and its variants are not generally universally Bayes consistent, except under additional structural assumptions, such as an inner product, a norm, finite doubling dimension, or a Besicovitch-type property.The metric spaces in which universal Bayes consistency is possible are the "essentially separable" ones -a new notion that we define, which is more general than standard separability. The existence of metric spaces that are not essentially separable is independent of the ZFC axioms of set theory. We prove that essential separability exactly characterizes the existence of a universal Bayes-consistent learner for the given metric space. In particular, this yields the first impossibility result for universal Bayes consistency.Taken together, these positive and negative results resolve the open problems posed in Kontorovich, Sabato, Weiss (2017).

show abstract

On Competitiveness of Nearest-Neighbor-Based Music Classification: A Methodological Critique

Cited by 6 publications

References 15 publications

Web-Based Music Genre Classification for Timeline Song Visualization and Analysis

Web-Based Music Genre Classification for Timeline Song Visualization and Analysis

Machine learning for music genre: multifaceted review and experimentation with audioset

Universal Bayes consistency in metric spaces

Contact Info

Product

Resources

About