Detection and Classification of Whale Acoustic Signals

This dissertation focuses on two vital challenges in relation to whale acoustic signals: detection and classification.

In detection, we evaluated the influence of the uncertain ocean environment on the spectrogram-based detector, and derived the likelihood ratio of the proposed Short Time Fourier Transform detector. Experimental results showed that the proposed detector outperforms detectors based on the spectrogram. The proposed detector is more sensitive to environmental changes because it includes phase information.

In classification, our focus is on finding a robust and sparse representation of whale vocalizations. Because whale vocalizations can be modeled as polynomial phase signals, we can represent the whale calls by their polynomial phase coefficients. In this dissertation, we used the Weyl transform to capture chirp rate information, and used a two dimensional feature set to represent whale vocalizations globally. Experimental results showed that our Weyl feature set outperforms chirplet coefficients and MFCC (Mel Frequency Cepstral Coefficients) when applied to our collected data.

Since whale vocalizations can be represented by polynomial phase coefficients, it is plausible that the signals lie on a manifold parameterized by these coefficients. We also studied the intrinsic structure of high dimensional whale data by exploiting its geometry. Experimental results showed that nonlinear mappings such as Laplacian Eigenmap and ISOMAP outperform linear mappings such as PCA and MDS, suggesting that the whale acoustic data is nonlinear.

We also explored deep learning algorithms on whale acoustic data. We built each layer as convolutions with either a PCA filter bank (PCANet) or a DCT filter bank (DCTNet). With the DCT filter bank, each layer has different a time-frequency scale representation, and from this, one can extract different physical information. Experimental results showed that our PCANet and DCTNet achieve high classification rate on the whale vocalization data set. The word error rate of the DCTNet feature is similar to the MFSC in speech recognition tasks, suggesting that the convolutional network is able to reveal acoustic content of speech signals.

Contributors to and impact of residual shunting after device closure of atrial septal defects.

BACKGROUND: The prevalence of residual shunt in patients after device closure of atrial septal defect and its impact on long-term outcome has not been previously defined. METHODS: From a prospective, single-institution registry of 408 patients, we selected individuals with agitated saline studies performed 1 year after closure. Baseline echocardiographic, invasive hemodynamic, and comorbidity data were compared to identify contributors to residual shunt. Survival was determined by review of the medical records and the Social Security Death Index. Survival analysis according to shunt included construction of Kaplan-Meier curves and Cox proportional hazards modeling. RESULTS: Among 213 analyzed patients, 27% were men and age at repair was 47 ± 17 years. Thirty patients (14%) had residual shunt at 1 year. Residual shunt was more common with Helex (22%) and CardioSEAL/STARFlex (40%) occluder devices than Amplatzer devices (9%; P = .005). Residual shunts were more common in whites (79% vs 46%, P = .004). At 7.3 ± 3.3 years of follow-up, 13 (6%) of patients had died, including 8 (5%) with Amplatzer, 5 (25%) with CardioSEAL/STARFlex, and 0 with Helex devices. Patients with residual shunting had a higher hazard of death (20% vs 4%, P = .001; hazard ratio 4.95 [1.59-14.90]). In an exploratory multivariable analysis, residual shunting, age, hypertension, coronary artery disease, and diastolic dysfunction were associated with death. CONCLUSIONS: Residual shunt after atrial septal defect device closure is common and adversely impacts long-term survival.