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.

Learning from Incredible Commitments: Evolution and Impact of Bilateral Investment Treaties

Ostensibly, BITs are the ideal international treaty. First, until just recently, they almost uniformly came with explicit dispute resolution mechanisms through which countries could face real costs for violation (Montt 2009). Second, the signing, ratification, and violation of them are easily accessible public knowledge. Thus countries presumably would face reputational costs for violating these agreements. Yet, these compliance devices have not dissuaded states from violating these agreements. Even more interestingly, in recent years, both developed and developing countries have moved towards modifying the investor-friendly provisions of these agreements. These deviations from the expectations of the credible commitment argument raise important questions about the field's assumptions regarding the ability of international treaties with commitment devices to effectively constrain state behavior.

Potential impacts of leakage from deep CO2 geosequestration on overlying freshwater aquifers.

Carbon Capture and Storage may use deep saline aquifers for CO(2) sequestration, but small CO(2) leakage could pose a risk to overlying fresh groundwater. We performed laboratory incubations of CO(2) infiltration under oxidizing conditions for >300 days on samples from four freshwater aquifers to 1) understand how CO(2) leakage affects freshwater quality; 2) develop selection criteria for deep sequestration sites based on inorganic metal contamination caused by CO(2) leaks to shallow aquifers; and 3) identify geochemical signatures for early detection criteria. After exposure to CO(2), water pH declines of 1-2 units were apparent in all aquifer samples. CO(2) caused concentrations of the alkali and alkaline earths and manganese, cobalt, nickel, and iron to increase by more than 2 orders of magnitude. Potentially dangerous uranium and barium increased throughout the entire experiment in some samples. Solid-phase metal mobility, carbonate buffering capacity, and redox state in the shallow overlying aquifers influence the impact of CO(2) leakage and should be considered when selecting deep geosequestration sites. Manganese, iron, calcium, and pH could be used as geochemical markers of a CO(2) leak, as their concentrations increase within 2 weeks of exposure to CO(2).

Learning from Geometry

Subspaces and manifolds are two powerful models for high dimensional signals. Subspaces model linear correlation and are a good fit to signals generated by physical systems, such as frontal images of human faces and multiple sources impinging at an antenna array. Manifolds model sources that are not linearly correlated, but where signals are determined by a small number of parameters. Examples are images of human faces under different poses or expressions, and handwritten digits with varying styles. However, there will always be some degree of model mismatch between the subspace or manifold model and the true statistics of the source. This dissertation exploits subspace and manifold models as prior information in various signal processing and machine learning tasks.

A near-low-rank Gaussian mixture model measures proximity to a union of linear or affine subspaces. This simple model can effectively capture the signal distribution when each class is near a subspace. This dissertation studies how the pairwise geometry between these subspaces affects classification performance. When model mismatch is vanishingly small, the probability of misclassification is determined by the product of the sines of the principal angles between subspaces. When the model mismatch is more significant, the probability of misclassification is determined by the sum of the squares of the sines of the principal angles. Reliability of classification is derived in terms of the distribution of signal energy across principal vectors. Larger principal angles lead to smaller classification error, motivating a linear transform that optimizes principal angles. This linear transformation, termed TRAIT, also preserves some specific features in each class, being complementary to a recently developed Low Rank Transform (LRT). Moreover, when the model mismatch is more significant, TRAIT shows superior performance compared to LRT.

The manifold model enforces a constraint on the freedom of data variation. Learning features that are robust to data variation is very important, especially when the size of the training set is small. A learning machine with large numbers of parameters, e.g., deep neural network, can well describe a very complicated data distribution. However, it is also more likely to be sensitive to small perturbations of the data, and to suffer from suffer from degraded performance when generalizing to unseen (test) data.

From the perspective of complexity of function classes, such a learning machine has a huge capacity (complexity), which tends to overfit. The manifold model provides us with a way of regularizing the learning machine, so as to reduce the generalization error, therefore mitigate overfiting. Two different overfiting-preventing approaches are proposed, one from the perspective of data variation, the other from capacity/complexity control. In the first approach, the learning machine is encouraged to make decisions that vary smoothly for data points in local neighborhoods on the manifold. In the second approach, a graph adjacency matrix is derived for the manifold, and the learned features are encouraged to be aligned with the principal components of this adjacency matrix. Experimental results on benchmark datasets are demonstrated, showing an obvious advantage of the proposed approaches when the training set is small.

Stochastic optimization makes it possible to track a slowly varying subspace underlying streaming data. By approximating local neighborhoods using affine subspaces, a slowly varying manifold can be efficiently tracked as well, even with corrupted and noisy data. The more the local neighborhoods, the better the approximation, but the higher the computational complexity. A multiscale approximation scheme is proposed, where the local approximating subspaces are organized in a tree structure. Splitting and merging of the tree nodes then allows efficient control of the number of neighbourhoods. Deviation (of each datum) from the learned model is estimated, yielding a series of statistics for anomaly detection. This framework extends the classical {\em changepoint detection} technique, which only works for one dimensional signals. Simulations and experiments highlight the robustness and efficacy of the proposed approach in detecting an abrupt change in an otherwise slowly varying low-dimensional manifold.

Test-pattern selection for screening small-delay defects in very-deep submicrometer integrated circuits

Timing-related defects are major contributors to test escapes and in-field reliability problems for very-deep submicrometer integrated circuits. Small delay variations induced by crosstalk, process variations, power-supply noise, as well as resistive opens and shorts can potentially cause timing failures in a design, thereby leading to quality and reliability concerns. We present a test-grading technique that uses the method of output deviations for screening small-delay defects (SDDs). A new gate-delay defect probability measure is defined to model delay variations for nanometer technologies. The proposed technique intelligently selects the best set of patterns for SDD detection from an n-detect pattern set generated using timing-unaware automatic test-pattern generation (ATPG). It offers significantly lower computational complexity and excites a larger number of long paths compared to a current generation commercial timing-aware ATPG tool. Our results also show that, for the same pattern count, the selected patterns provide more effective coverage ramp-up than timing-aware ATPG and a recent pattern-selection method for random SDDs potentially caused by resistive shorts, resistive opens, and process variations. © 2010 IEEE.

From offshore to onshore: multiple origins of shallow-water corals from deep-sea ancestors.

Shallow-water tropical reefs and the deep sea represent the two most diverse marine environments. Understanding the origin and diversification of this biodiversity is a major quest in ecology and evolution. The most prominent and well-supported explanation, articulated since the first explorations of the deep sea, holds that benthic marine fauna originated in shallow, onshore environments, and diversified into deeper waters. In contrast, evidence that groups of marine organisms originated in the deep sea is limited, and the possibility that deep-water taxa have contributed to the formation of shallow-water communities remains untested with phylogenetic methods. Here we show that stylasterid corals (Cnidaria: Hydrozoa: Stylasteridae)--the second most diverse group of hard corals--originated and diversified extensively in the deep sea, and subsequently invaded shallow waters. Our phylogenetic results show that deep-water stylasterid corals have invaded the shallow-water tropics three times, with one additional invasion of the shallow-water temperate zone. Our results also show that anti-predatory innovations arose in the deep sea, but were not involved in the shallow-water invasions. These findings are the first robust evidence that an important group of tropical shallow-water marine animals evolved from deep-water ancestors.

Transmission of single HIV-1 genomes and dynamics of early immune escape revealed by ultra-deep sequencing.

We used ultra-deep sequencing to obtain tens of thousands of HIV-1 sequences from regions targeted by CD8+ T lymphocytes from longitudinal samples from three acutely infected subjects, and modeled viral evolution during the critical first weeks of infection. Previous studies suggested that a single virus established productive infection, but these conclusions were tempered because of limited sampling; now, we have greatly increased our confidence in this observation through modeling the observed earliest sample diversity based on vastly more extensive sampling. Conventional sequencing of HIV-1 from acute/early infection has shown different patterns of escape at different epitopes; we investigated the earliest escapes in exquisite detail. Over 3-6 weeks, ultradeep sequencing revealed that the virus explored an extraordinary array of potential escape routes in the process of evading the earliest CD8 T-lymphocyte responses--using 454 sequencing, we identified over 50 variant forms of each targeted epitope during early immune escape, while only 2-7 variants were detected in the same samples via conventional sequencing. In contrast to the diversity seen within epitopes, non-epitope regions, including the Envelope V3 region, which was sequenced as a control in each subject, displayed very low levels of variation. In early infection, in the regions sequenced, the consensus forms did not have a fitness advantage large enough to trigger reversion to consensus amino acids in the absence of immune pressure. In one subject, a genetic bottleneck was observed, with extensive diversity at the second time point narrowing to two dominant escape forms by the third time point, all within two months of infection. Traces of immune escape were observed in the earliest samples, suggesting that immune pressure is present and effective earlier than previously reported; quantifying the loss rate of the founder virus suggests a direct role for CD8 T-lymphocyte responses in viral containment after peak viremia. Dramatic shifts in the frequencies of epitope variants during the first weeks of infection revealed a complex interplay between viral fitness and immune escape.

Beam-target double-spin asymmetry A{LT} in charged pion production from deep inelastic scattering on a transversely polarized {3}He target at 1.4

We report the first measurement of the double-spin asymmetry A{LT} for charged pion electroproduction in semi-inclusive deep-inelastic electron scattering on a transversely polarized {3}He target. The kinematics focused on the valence quark region, 0.16

The impact of geologic variability on capacity and cost estimates for storing CO 2 in deep-saline aquifers

While numerous studies find that deep-saline sandstone aquifers in the United States could store many decades worth of the nation's current annual CO 2 emissions, the likely cost of this storage (i.e. the cost of storage only and not capture and transport costs) has been harder to constrain. We use publicly available data of key reservoir properties to produce geo-referenced rasters of estimated storage capacity and cost for regions within 15 deep-saline sandstone aquifers in the United States. The rasters reveal the reservoir quality of these aquifers to be so variable that the cost estimates for storage span three orders of magnitude and average>$100/tonne CO 2. However, when the cost and corresponding capacity estimates in the rasters are assembled into a marginal abatement cost curve (MACC), we find that ~75% of the estimated storage capacity could be available for<$2/tonne. Furthermore, ~80% of the total estimated storage capacity in the rasters is concentrated within just two of the aquifers-the Frio Formation along the Texas Gulf Coast, and the Mt. Simon Formation in the Michigan Basin, which together make up only ~20% of the areas analyzed. While our assessment is not comprehensive, the results suggest there should be an abundance of low-cost storage for CO 2 in deep-saline aquifers, but a majority of this storage is likely to be concentrated within specific regions of a smaller number of these aquifers. © 2011 Elsevier B.V.

Biogeography in deep time - What do phylogenetics, geology, and paleoclimate tell us about early platyrrhine evolution?

Molecular data have converged on a consensus about the genus-level phylogeny of extant platyrrhine monkeys, but for most extinct taxa and certainly for those older than the Pleistocene we must rely upon morphological evidence from fossils. This raises the question as to how well anatomical data mirror molecular phylogenies and how best to deal with discrepancies between the molecular and morphological data as we seek to extend our phylogenies to the placement of fossil taxa. Here I present parsimony-based phylogenetic analyses of extant and fossil platyrrhines based on an anatomical dataset of 399 dental characters and osteological features of the cranium and postcranium. I sample 16 extant taxa (one from each platyrrhine genus) and 20 extinct taxa of platyrrhines. The tree structure is constrained with a "molecular scaffold" of extant species as implemented in maximum parsimony using PAUP with the molecular-based 'backbone' approach. The data set encompasses most of the known extinct species of platyrrhines, ranging in age from latest Oligocene (∼26 Ma) to the Recent. The tree is rooted with extant catarrhines, and Late Eocene and Early Oligocene African anthropoids. Among the more interesting patterns to emerge are: (1) known early platyrrhines from the Late Oligocene through Early Miocene (26-16.5Ma) represent only stem platyrrhine taxa; (2) representatives of the three living platyrrhine families first occur between 15.7 Ma and 13.5 Ma; and (3) recently extinct primates from the Greater Antilles (Cuba, Jamaica, Hispaniola) are sister to the clade of extant platyrrhines and may have diverged in the Early Miocene. It is probable that the crown platyrrhine clade did not originate before about 20-24 Ma, a conclusion consistent with the phylogenetic analysis of fossil taxa presented here and with recent molecular clock estimates. The following biogeographic scenario is consistent with the phylogenetic findings and climatic and geologic evidence: Tropical South America has been a center for platyrrhine diversification since platyrrhines arrived on the continent in the middle Cenozoic. Platyrrhines dispersed from tropical South America to Patagonia at ∼25-24 Ma via a "Paraná Portal" through eastern South America across a retreating Paranense Sea. Phylogenetic bracketing suggests Antillean primates arrived via a sweepstakes route or island chain from northern South America in the Early Miocene, not via a proposed land bridge or island chain (GAARlandia) in the Early Oligocene (∼34 Ma). Patagonian and Antillean platyrrhines went extinct without leaving living descendants, the former at the end of the Early Miocene and the latter within the past six thousand years. Molecular evidence suggests crown platyrrhines arrived in Central America by crossing an intermittent connection through the Isthmus of Panama at or after 3.5Ma. Any more ancient Central American primates, should they be discovered, are unlikely to have given rise to the extant Central American taxa in situ.