Automatic Feature Extraction for Classifying Audio Data

Mierswa, Ingo; Morik, Katharina

doi:10.1007/s10994-005-5824-7

Cited by 126 publications

(58 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…We believe that these features, extracted over these window lengths, are diverse enough to capture important historical information from the signals to be used in models. There are other features that could be extracted, or features can be extracted using automated and supervised methods (Guo, Jack, & Nandi, 2005;Mierswa & Morik, 2005;Hamel & Eck, 2010). …”

Section: Temporal Feature Extractionmentioning

confidence: 99%

Investigating the Feasibility of Vehicle Telemetry Data as a Means of Predicting Driver Workload

Taylor

Griffiths

Bhalerao

et al. 2017

International Journal of Mobile Human Computer Interaction

View full text Add to dashboard Cite

Driving is a safety critical task that requires a high level of attention and workload from the driver. Despite this, people often also perform secondary tasks such as eating or using a mobile phone, which increase workload levels and divert cognitive and physical attention from the primary task of driving. If a vehicle is aware that the driver is currently under high workload, the vehicle functionality can be changed in order to minimize any further demand. Traditionally, workload measurements have been performed using intrusive means such as physiological sensors. Another approach may be to use vehicle telemetry data as a performance measure for workload. In this paper, we present the Warwick-JLR Driver Monitoring Dataset (DMD) and analyse it to investigate the feasibility of using vehicle telemetry data for determining the driver workload. We perform a statistical analysis of subjective ratings, physiological data, and vehicle telemetry data collected during a track study. A data mining methodology is then presented to build predictive models using this data, for the driver workload monitoring problem

show abstract

Section: Temporal Feature Extractionmentioning

confidence: 99%

Investigating the Feasibility of Vehicle Telemetry Data as a Means of Predicting Driver Workload

Taylor

Griffiths

Bhalerao

et al. 2017

International Journal of Mobile Human Computer Interaction

View full text Add to dashboard Cite

show abstract

“…The reason was that the method shall only use basic methods thus Fast Fourier Transformation has been used. Elementary feature extraction operations can be executed in any order and allow the creation of a set of feature extraction operations the can be different for each problem [21]. This makes elementary extraction operations also good for machine learning.…”

Section: Feature Extraction and Sensor Fusionmentioning

confidence: 99%

Decision trees and the eﬀects of feature extraction parameters for robust sensor network design

Gerdes

Galar

Scholz

2016

EiN

View full text Add to dashboard Cite

IntroductionThis paper presents a condition monitoring system with sensor optimization capabilities to prevent unscheduled delays in the aircraft industry. Unscheduled delays cost airlines a great deal of money but can be prevented by condition monitoring [11]. The aim was to develop a simple condition monitoring system that can be understood by humans and modified by experts to incorporate knowledge that is not in the learning data set, using decision trees as the main tool. Decision Gerdes M, GAlAr d, scholz d. decision trees and the effects of feature extraction parameters for robust sensor network design. eksploatacja i Niezawodnosc - Maintenance and reliability 2017; 19 (1): 31-42, http://dx.doi.org/10.17531/ein.2017.1.5. trees satisfy the requirements and provide a ranking of data sources for condition monitoring.The first section of the paper gives the motivation for developing a condition monitoring system with sensor optimization capabilities and explains the basic concepts of the proposed method. The second section explains the method in detail. Section three discusses the experiments validating it. The results of the validation experiments are given in section four. The paper concludes with a discussion of the results. New and better monitoring approaches are required for condition monitoring, because systems are becoming more complex and more difficult to monitor [32]. Condition monitoring requires reliable sensors. To obtain enough sensing data, special attention should be given to optimizing sensor allocation to ensure system diagnosability, lower sensing cost and reduce time to diagnosis [37]. Sensors can be used to determine the system health of control systems; a failed sensor can lead to a loss of process control [18]. The information about a system is incomplete, if a sensor fails. Complex systems are often monitored by multiple sensors. An advantage of a multi sensor system is that a single failed sensor shows its effects in multiple sensors [18].This means the system condition is defined by all information from the sensors. However, the system's health status becomes uncertain when a sensor fails or sends wrong data. This could trigger incorrect maintenance, including maintenance on a part with no failure, as well as long maintenance times to find the correct fault or not noticing the fault at all.The Safety Integrity Level (SIL) defines the probability that the system safety function for a Safety Instrumented System (SIS) can be executed. There are four SILs; level four is the level with the highest probability that an SIS can be performed. Sensor failure detection (sensor validation) is a critical part of the safety function of a system. When a failure is detected SIS is put into a safe state to avoid risk and damage to humans and machines [14,13].Redundancy is used to reduce the risk of model uncertainty [5]. One way to create sensor redundancy is hardware redundancy; another is analytical redundancy [5]. Analytical redundancy assumes multiple sensors deliver the same information, and,...

show abstract

“…It was shown that some of the constructed features could indeed improve music classification performance relative to conventional features. Another such example was presented by Mierswa and Morik [25], where method trees consisting of ad hoc features for a given audio signal were introduced. The trees were automatically generated with GP by combining elementary feature extraction methods.…”

Section: Related Workmentioning

confidence: 99%

An evolutionary feature synthesis approach for content-based audio retrieval

Mäkinen

Kıranyaz

Raitoharju

et al. 2012

J AUDIO SPEECH MUSIC PROC.

View full text Add to dashboard Cite

A vast amount of audio features have been proposed in the literature to characterize the content of audio signals. In order to overcome specific problems related to the existing features (such as lack of discriminative power), as well as to reduce the need for manual feature selection, in this article, we propose an evolutionary feature synthesis technique with a built-in feature selection scheme. The proposed synthesis process searches for optimal linear/ nonlinear operators and feature weights from a pre-defined multi-dimensional search space to generate a highly discriminative set of new (artificial) features. The evolutionary search process is based on a stochastic optimization approach in which a multi-dimensional particle swarm optimization algorithm, along with fractional global best formation and heterogeneous particle behavior techniques, is applied. Unlike many existing feature generation approaches, the dimensionality of the synthesized feature vector is also searched and optimized within a set range in order to better meet the varying requirements set by many practical applications and classifiers. The new features generated by the proposed synthesis approach are compared with typical low-level audio features in several classification and retrieval tasks. The results demonstrate a clear improvement of up to 15-20% in average retrieval performance. Moreover, the proposed synthesis technique surpasses the synthesis performance of evolutionary artificial neural networks, exhibiting a considerable capability to accurately distinguish among different audio classes.

show abstract

Automatic Feature Extraction for Classifying Audio Data

Cited by 126 publications

References 19 publications

Investigating the Feasibility of Vehicle Telemetry Data as a Means of Predicting Driver Workload

Investigating the Feasibility of Vehicle Telemetry Data as a Means of Predicting Driver Workload

Decision trees and the eﬀects of feature extraction parameters for robust sensor network design

An evolutionary feature synthesis approach for content-based audio retrieval

Contact Info

Product

Resources

About