Multiresolution classification with semi-supervised learning for indirect bridge structural health monitoring

2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Sandryhaila

Lederman

et al. 2014

Self Cite

We propose a novel recovery algorithm for signals with complex, irregular structure that is commonly represented by graphs. Our approach is a generalization of the signal inpainting technique from classical signal processing. We formulate corresponding minimization problems and demonstrate that in many cases they have closed-form solutions. We discuss a relation of the proposed approach to regression, provide an upper bound on the error for our algorithm and compare the proposed technique with other existing algorithms on realworld datasets.

Section: Bridge Condition Identificationmentioning

confidence: 99%

Signal inpainting on graphs via total variation minimization

2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Sandryhaila

Lederman

et al. 2014

Self Cite

“…We capture vibration characteristics of the bridge from the vibration of the traversing vehicle through the acceleration signal. We collected 30 acceleration signals for each of 13 [8], [12]. Experimental Setup.…”

Section: B Damage Detection In Bridge Structure Monitoringmentioning

confidence: 99%

“…The optimal filter coefficients are designed to fit the confidence vectors of the labeled signals from all the graph shifts to the ground truth and to minimize the labeling uncertainty of the unlabeled signals. We introduce the labeling uncertainty measure [8], [9] as…”

Section: Adaptive Graph Filteringmentioning

confidence: 99%

Adaptive graph filtering: Multiresolution classification on graphs

2013 IEEE Global Conference on Signal and Information Processing

Sandryhaila

Moura

et al. 2013

Self Cite

Abstract-We present an adaptive graph filtering approach to semi-supervised classification. Adaptive graph filters combine decisions from multiple graph filters using a weighting function that is optimized in a semi-supervised manner. We also demonstrate the multiresolution property of adaptive graph filters by connecting them to the diffusion wavelets. In our experiments, we apply the adaptive graph filters to the classification of online blogs and damage identification in indirect bridge structural health monitoring.

“…We thus propose a novel classification framework that takes advantage of multiresolution classification [4], [11], which extracts hidden features in localized time-frequency regions (subbands), and semi-supervised learning [12], which uses both labeled and unlabeled signals for classification; we make this possible by developing a semi-supervised weighting algorithm. In the new framework, (1) each localized subband contributes to the classification by its discriminative power; and (2) both labeled and unlabeled signals provide information.…”

mentioning

confidence: 99%

Semi-Supervised Multiresolution Classification Using Adaptive Graph Filtering With Application to Indirect Bridge Structural Health Monitoring

IEEE Trans. Signal Process.

Cerda

Rizzo

et al. 2014

Self Cite

125

We present a multiresolution classification framework with semi-supervised learning on graphs with application to the indirect bridge structural health monitoring. Classification in real-world applications faces two main challenges: reliable features can be hard to extract and few labeled signals are available for training. We propose a novel classification framework to address these problems: we use a multiresolution framework to deal with nonstationarities in the signals and extract features in each localized time-frequency region and semi-supervised learning to train on both labeled and unlabeled signals. We further propose an adaptive graph filter for semi-supervised classification that allows for classifying unlabeled as well as unseen signals and for correcting mislabeled signals. We validate the proposed framework on indirect bridge structural health monitoring and show that it performs significantly better than previous approaches.Index Terms-Multiresolution classification, semi-supervised learning, discrete signal processing on graphs, adaptive graph filter, indirect bridge structural health monitoring.