Continuous HMM-Based Seismic-Event Classification at Deception Island, Antarctica

The Automated Identification of Volcanic Earthquakes: Concepts, Applications and Challenges

“…Additional features can be extracted from spectral representations by measuring morphological attributes such as the mean frequencies of the five highest peaks, energies in given frequency bands (Curilem et al, 2009;Romeo, 1994;Romeo et al, 1995) and the instantaneous frequency , or by computing variables such as the Mel-frequency cepstral coefficients (MFCCs), their associated log-energies and the so-called delta and delta-delta coefficients (Benítez et al, 2007;Gutiérrez et al, 2009;. Spectra and spectrograms are typically computed by using the Fourier or the cosine transforms.…”

“…Publication Number of states (Benítez et al, 2007) 11 - ) 3,4,5 (Alasonati et al, 2006 2 (Gutiérrez et al, 2006) 13,20 13 (Hoogenboezem, 2010) 20 (Ibáñez et al, 2009) 17,22 (Ohrnberger, 2001) 6,12 Table 3. Configurations of HMMs applied in the reviewed publications.…”

Selection of time-variant features for earthquake classification at the Nevado-del-Ruiz volcano

“…Taking into account the seismogram variability over time, time-variant features have been used to extract information, such as, linear frequency cepstral coefficients (LFCC) (Benítez et al, 2007), mel-frequency cepstral coefficients (MFCC) , instantaneous bandwidth, instantaneous frequency, seismic envelope, the centroid time , and the half octave band (HOB) . Throughout this study, instead of a longterm parameter set extracted from the seismogram, the volcano state events are detected by using short-time features, where each seismogram frame is quantified by a single vector.…”

Automatic Classification of Volcanic Earthquakes in HMM-Induced Vector Spaces

“…Several feature extraction 1 and classification approaches have been tested in previous studies [2]. Among the classification approaches, hidden Markov models (HMMs) have proved to be convenient classification tools for the sequential and unequal-length nature of the seismic signals [3][4][5] (conference papers and other older publications are not cited here due to space constraints, refer again to [2] for additional references). Even though these studies show the usefulness of such approaches, the full potentialities of HMM-based classification systems have not been completely exploited.…”