Outlier detection and robust regression for correlated data

Yuen, Ka‐Veng; Ortiz, Gilberto A.

doi:10.1016/j.cma.2016.10.004

Cited by 46 publications

(42 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Note that in the proposed hierarchical outlier detection approach, the probability threshold is 0.5 for identifying both local and global outliers. It is intuitive because an outlier probability of 0.5 implies that this data point has equal probability of being an outlier or a regular data point 49–51 …”

Section: Proposed Hierarchical Outlier Detection Approachmentioning

confidence: 99%

See 1 more Smart Citation

Hierarchical outlier detection approach for online distributed structural identification

Huang

Yuen

2020

Struct Control Health Monit

View full text Add to dashboard Cite

In this paper, a hierarchical outlier detection approach is proposed for online distributed structural identification. In contrast to centralized identification, distributed identification extracts important features from the raw response data at the sensor nodes and transmits only them to the base station. Therefore, outlier detection is substantially more complicated than the traditional approach. In the proposed method, the local outliers in the raw data are detected directly at the corresponding sensor node, and they are excluded from further processing. However, if a sensor node is biased or exhibits other patterned outliers, these outliers will be undetectable at the sensor node level. It is necessary to conduct another level of outlier detection at the base station, namely, global outlier detection, before fusion. These two levels of outlier detection are of different nature. Local outlier detection concerns directly with the raw response data, whereas the targets of global outlier detection are the local estimation results of the stiffness parameters. Therefore, they require different mathematical tools. The proposed hierarchical outlier detection approach detects the local outliers according to the outlier probability of the data points at the sensor nodes, whereas it detects the global outliers according to the outlier probability of the local estimation results. By excluding both types of outliers, reliable online distributed structural identification can be achieved. Two examples are presented to demonstrate the proposed method.

show abstract

Section: Proposed Hierarchical Outlier Detection Approachmentioning

confidence: 99%

“…One of the advantages for such definition is that it can avoid the subjectivity in the prescribed threshold value, in contrast to most existing outlier detection techniques. This method was extended to problems with correlated noise 50 and centralized structural identification 51,52 …”

Section: Introductionmentioning

confidence: 99%

Hierarchical outlier detection approach for online distributed structural identification

Huang

Yuen

2020

Struct Control Health Monit

View full text Add to dashboard Cite

show abstract

“…Outlier detection has attracted a significant interest in several different areas. The authors in [ 17 ] applied outlier detection to estimate peak ground accelerations in seismic data. An algorithm for collision and hazard detection for motorcycles via inertial measurements based on outlier detection was presented in [ 18 ].…”

Section: Related Workmentioning

confidence: 99%

Detecting Steps Walking at very Low Speeds Combining Outlier Detection, Transition Matrices and Autoencoders from Acceleration Patterns

Muñoz-Organero

Blázquez

2017

Sensors

View full text Add to dashboard Cite

In this paper, we develop and validate a new algorithm to detect steps while walking at speeds between 30 and 40 steps per minute based on the data sensed from a single tri-axial accelerometer. The algorithm concatenates three consecutive phases. First, an outlier detection is performed on the sensed data based on the Mahalanobis distance to pre-detect candidate points in the acceleration time series that may contain a ground contact segment of data while walking. Second, the acceleration segment around the pre-detected point is used to calculate the transition matrix in order to capture the time dependencies. Finally, autoencoders, trained with data segments containing ground contact transition matrices from acceleration series from labeled steps are used to reconstruct the computed transition matrices at each pre-detected point. A similarity index is used to assess if the pre-selected point contains a true step in the 30–40 steps per minute speed range. Our experimental results, based on a database from three different participants performing similar activities to the target one, are able to achieve a recall = 0.88 with precision = 0.50 improving the results when directly applying the autoencoders to acceleration patterns (recall = 0.77 with precision = 0.50).

show abstract

“…Bayesian inference has been gaining popularity over the last decades in mechanics and engineering problems, and this can be evidenced in the amount of applications found in various areas such as damage detection, [29][30][31][32][33] geotechnical engineering, [31][32][33][34] structural dynamics, [35][36][37][38][39][40][41] air quality prediction, [42] reliability analysis, [43][44][45][46] hydraulic engineering, [47] random vibrations, [48] and outlier detection. [49] However, if BNGR is directly applied to model updating using modal data, it will require a huge number of samples due to the complexity of the problem. This will be elaborated later.…”

Section: Introductionmentioning

confidence: 99%

Multiresolution Bayesian nonparametric general regression for structural model updating

Yuen

Ortiz

2017

Struct Control Health Monit

Self Cite

View full text Add to dashboard Cite

Summary A novel Bayesian method, namely, multiresolution Bayesian nonparametric general regression (MR‐BNGR), is proposed for structural model updating using modal data, that is, identified natural frequencies and mode shapes. In this method, the model updating problem is posed as a non‐linear regression problem from the modal data to the structural parameters. The proposed method is nonparametric, so it does not require an explicit functional form of this mapping. Instead, it utilizes the input–output data to adaptively model its relationship. Its multiresolution nature allows to zoom into the significant region in stages to search the optimal point. Furthermore, the estimation uncertainty can be quantified. Training of the MR‐BNGR network is very straightforward and computationally economical. Examples of a 20‐storey shear building and a three‐dimensional truss are provided to demonstrate the capabilities of the proposed MR‐BNGR method, and the results confirm the effectiveness of this novel method.

show abstract

Outlier detection and robust regression for correlated data

Cited by 46 publications

References 41 publications

Hierarchical outlier detection approach for online distributed structural identification

Hierarchical outlier detection approach for online distributed structural identification

Detecting Steps Walking at very Low Speeds Combining Outlier Detection, Transition Matrices and Autoencoders from Acceleration Patterns

Multiresolution Bayesian nonparametric general regression for structural model updating

Contact Info

Product

Resources

About