Description and Discussion on DCASE2020 Challenge Task2: Unsupervised Anomalous Sound Detection for Machine Condition Monitoring

Kawaguchi, Yohei; Imoto, Keisuke; Nakamura, Toshiki; Nikaido, Yuki; Tanabe, Ryo; Purohit, Harsh; Suefusa, Kaori; Endo, Takashi; Yasuda, Masahiro; Harada, Noboru

doi:10.48550/arxiv.2006.05822

Cited by 25 publications

(27 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the superior performance of classification-based methods on our application of interest [8]- [10], we focus our review on this class of methods and refer interested readers to comprehensive reviews [7], [17] for more information. A representative method is Deep Support Vector Data Description (SVDD) [6], a one-class classification method, where the neural network learns a representation that enforces the majority of normal samples to fall within a hypersphere.…”

Section: Related Workmentioning

confidence: 99%

“…Given its superior performance, especially on similar problems [8]- [10], we adopt a classification-based approach for anomaly detection [11], where auxiliary classification tasks are defined on the available meta-data. However, existing classification-based methods [11], [12] are not designed for anomaly detection under domain shift, and thus are not sufficient for competitive performance on their own, which we show in Section V. Thus, we propose a novel approach, with bi-level optimization, to tackle the challenging problem of anomaly detection under domain shifts.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Learning to Adapt to Domain Shifts with Few-shot Samples in Anomalous Sound Detection

Chen¹,

Bondi²,

Das³

2022

Preprint

View full text Add to dashboard Cite

Anomaly detection has many important applications, such as monitoring industrial equipment. Despite recent advances in anomaly detection with deep-learning methods, it is unclear how existing solutions would perform under out-ofdistribution scenarios, e.g., due to shifts in machine load or environmental noise. Grounded in the application of machine health monitoring, we propose a framework that adapts to new conditions with few-shot samples. Building upon prior work, we adopt a classification-based approach for anomaly detection and show its equivalence to mixture density estimation of the normal samples. We incorporate an episodic training procedure to match the few-shot setting during inference. We define multiple auxiliary classification tasks based on meta-information and leverage gradient-based meta-learning to improve generalization to different shifts. We evaluate our proposed method on a recently-released dataset of audio measurements from different machine types. It improved upon two baselines by around 10% and is on par with best-performing model reported on the dataset.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Learning to Adapt to Domain Shifts with Few-shot Samples in Anomalous Sound Detection

Chen¹,

Bondi²,

Das³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…In this paper, we focus on anomalous sound detection (ASD) [2] for cavitation or other faults in complex industrial mechanical systems, which has been an active research topic in industry during these years. ASD is the task of determining whether the acoustic signal emitted from a target machine is normal or abnormal, which is a fundamental technology in the fourth industrial revolution to monitor the health of machines by 'listening' to their acoustics [3].…”

Section: Introductionmentioning

confidence: 99%

Regional-Local Adversarially Learned One-Class Classifier Anomalous Sound Detection in Global Long-Term Space

Sha,

Faber,

Gou

et al. 2022

Preprint

View full text Add to dashboard Cite

Anomalous sound detection (ASD) is one of the most significant tasks of mechanical equipment monitoring and maintaining in complex industrial systems. In practice, it is vital to precisely identify abnormal status of the working mechanical system, which can further facilitate the failure troubleshooting. In this paper, we propose a multi-pattern adversarial learning one-class classification framework, which allows us to use both the generator and the discriminator of an adversarial model for efficient ASD. The core idea is learning to reconstruct the normal patterns of acoustic data through two different patterns of auto-encoding generators, which succeeds in extending the fundamental role of a discriminator from identifying real and fake data to distinguishing between regional and local pattern reconstructions. Furthermore, we present a global filter layer for long-term interactions in the frequency domain space, which directly learns from the original data without introducing any human priors. Extensive experiments performed on four real-world datasets from different industrial domains (three cavitation datasets provided by SAMSON AG, and one existing publicly) for anomaly detection show superior results, and outperform recent state-of-the-art ASD methods.

show abstract

“…Sound event detection (SED), in which the types of sound event are identified and their onset and offset in an audio recording are estimated, is one of the principal tasks in environmental sound analysis [1,2]. Recently, many works have addressed SED because it plays an important role in realizing various applications using artificial intelligence in sounds, e.g., automatic life logging, machine monitoring, automatic surveillance, media retrieval, and biomonitoring systems [3,4,5,6,7,8].…”

Section: Introductionmentioning

confidence: 99%

Impact of Sound Duration and Inactive Frames on Sound Event Detection Performance

Imoto

Mishima²,

Arai³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

In many methods of sound event detection (SED), a segmented time frame is regarded as one data sample to model training. The durations of sound events greatly depend on the sound event class, e.g., the sound event "fan" has a long duration, whereas the sound event "mouse clicking" is instantaneous. Thus, the difference in the duration between sound event classes results in a serious data imbalance in SED. Moreover, most sound events tend to occur occasionally; therefore, there are many more inactive time frames of sound events than active frames. This also causes a severe data imbalance between active and inactive frames. In this paper, we investigate the impact of sound duration and inactive frames on SED performance by introducing four loss functions, such as simple reweighting loss, inverse frequency loss, asymmetric focal loss, and focal batch Tversky loss. Then, we provide insights into how we tackle this imbalance problem.

show abstract

Description and Discussion on DCASE2020 Challenge Task2: Unsupervised Anomalous Sound Detection for Machine Condition Monitoring

Cited by 25 publications

References 29 publications

Learning to Adapt to Domain Shifts with Few-shot Samples in Anomalous Sound Detection

Learning to Adapt to Domain Shifts with Few-shot Samples in Anomalous Sound Detection

Regional-Local Adversarially Learned One-Class Classifier Anomalous Sound Detection in Global Long-Term Space

Impact of Sound Duration and Inactive Frames on Sound Event Detection Performance

Contact Info

Product

Resources

About