Acoustic Scene Classification: A Competition Review

Gharib, Shayan; Derrar, Honain; Niizumi, Daisuke; Senttula, Tuukka; Tommola, Janne; Heittola, Toni; Virtanen, Tuomas; Huttunen, Heikki

doi:10.1109/mlsp.2018.8517000

Cited by 12 publications

(13 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…As a result, the filtering operation to remove the harmful augmented samples performed in [ 31 , 33 ] is not necessary here. Gharib et al [ 36 ] applied a similar random erasing method for ASC and achieved an improvement of 0.13 percent compared with their baseline system.…”

Section: Related Workmentioning

confidence: 99%

Clustering by Errors: A Self-Organized Multitask Learning Method for Acoustic Scene Classification

Zheng

Zhao

2021

Sensors

View full text Add to dashboard Cite

Acoustic scene classification (ASC) tries to inference information about the environment using audio segments. The inter-class similarity is a significant issue in ASC as acoustic scenes with different labels may sound quite similar. In this paper, the similarity relations amongst scenes are correlated with the classification error. A class hierarchy construction method by using classification error is then proposed and integrated into a multitask learning framework. The experiments have shown that the proposed multitask learning method improves the performance of ASC. On the TUT Acoustic Scene 2017 dataset, we obtain the ensemble fine-grained accuracy of 81.4%, which is better than the state-of-the-art. By using multitask learning, the basic Convolutional Neural Network (CNN) model can be improved by about 2.0 to 3.5 percent according to different spectrograms. The coarse category accuracies (for two to six super-classes) range from 77.0% to 96.2% by single models. On the revised version of the LITIS Rouen dataset, we achieve the ensemble fine-grained accuracy of 83.9%. The multitask learning models obtain an improvement of 1.6% to 1.8% compared to their basic models. The coarse category accuracies range from 94.9% to 97.9% for two to six super-classes with single models.

show abstract

Section: Related Workmentioning

confidence: 99%

Clustering by Errors: A Self-Organized Multitask Learning Method for Acoustic Scene Classification

Zheng

Zhao

2021

Sensors

View full text Add to dashboard Cite

show abstract

“…Para melhorar a generalização de um modelo de aprendizagem, o ideal é treiná-lo com a maior quantidade de dados possível [19]. Entretanto, a quantidade de dados disponível No presente trabalho, é utilizada a técnica mixup data augmentation [20], também usada em [21] e [22]. No mixup, exemplos virtuais são gerados a partir da interpolação linear entre dois pares exemplo/rótulo amostrados aleatoriamente do conjunto de treinamento.…”

Section: Data Augmentationunclassified

“…Tal processo consiste no ensemble [25] dos modelos obtidos na validação cruzada, onde a saída é a média aritmética da predição dos 5 modelos. A integração dos métodos de ensemble com a validação cruzada k-fold é sugerida por [21], que observou melhorias de desempenho significativas ao utilizar tal técnica.…”

Section: Treinamento E Aplicaçãounclassified

Classificação de Cenas Acústicas Utilizando Técnicas de Aprendizagem Profunda

Zanco¹

2019

Anais De XXXVII Simpósio Brasileiro De Telecomunicações E Processamento De Sinais

View full text Add to dashboard Cite

Resumo-Neste artigo são apresentadas contribuições que visam aumentar o desempenho do classificador de cenas acústicas baseline proposto no contexto do Detection and Classification of Acoustic Scenes and Events 2018 (DCASE2018) Challenge. Tais contribuições consistem em alterações na estrutura da rede neural convolucional como também na utilização de estratégias de data augmentation e ensemble de modelos. Os resultados obtidos mostram uma acurácia de 72,04% para o conjunto de desenvolvimento e 68,5% para o conjunto de avaliação, significativamente superiores às do baseline, que é de 59,7% e 61,0%, respectivamente.

show abstract

“…The historical preview of previous research and general framework for ASC can be found in [1], and an overview of current methods based on deep learning can be found in [2]. The ASC is one of the main tasks of the DCASE challenge 1 , and reviews of the most recent competitions can be found in [3], [4]. In this paper, we focus on the specific problem of mismatched domains (also known as the domain shift), where the mismatch is primarily caused by the usage of different recording devices.…”

Section: Introductionmentioning

confidence: 99%

Adversarial Domain Adaptation with Paired Examples for Acoustic Scene Classification on Different Recording Devices

Kacprzak¹,

Kowalczyk²

2021

Preprint

View full text Add to dashboard Cite

In classification tasks, the classification accuracy diminishes when the data is gathered in different domains.To address this problem, in this paper, we investigate several adversarial models for domain adaptation (DA) and their effect on the acoustic scene classification task. The studied models include several types of generative adversarial networks (GAN), with different loss functions, and the so-called cycle GAN which consists of two interconnected GAN models. The experiments are performed on the DCASE20 challenge task 1A dataset, in which we can leverage the paired examples of data recorded using different devices, i.e., the source and target domain recordings. The results of performed experiments indicate that the best performing domain adaptation can be obtained using the cycle GAN, which achieves as much as 66% relative improvement in accuracy for the target domain device, while only 6% relative decrease in accuracy on the source domain. In addition, by utilizing the paired data examples, we are able to improve the overall accuracy over the model trained using larger unpaired data set, while decreasing the computational cost of the model training.

show abstract

Acoustic Scene Classification: A Competition Review

Cited by 12 publications

References 20 publications

Clustering by Errors: A Self-Organized Multitask Learning Method for Acoustic Scene Classification

Clustering by Errors: A Self-Organized Multitask Learning Method for Acoustic Scene Classification

Classificação de Cenas Acústicas Utilizando Técnicas de Aprendizagem Profunda

Adversarial Domain Adaptation with Paired Examples for Acoustic Scene Classification on Different Recording Devices

Contact Info

Product

Resources

About