Automatic Audio Segmentation Using the Generalized Likelihood Ratio

This paper presents a novel technique for segmenting an audio stream into homogeneous regions according to speaker identities, background noise, music, environmental and channel conditions. Audio segmentation is useful in audio diarization systems, which aim to annotate an input audio stream with information that attributes temporal regions of the audio into their specific sources. The segmentation method introduced in this paper is performed using the Generalized Likelihood Ratio (GLR), computed between two adjacent sliding windows over preprocessed speech. This approach is inspired by the popular segmentation method proposed by the pioneering work of Chen and Gopalakrishnan, using the Bayesian Information Criterion (BIC) with an expanding search window. This paper will aim to identify and address the shortcomings associated with such an approach. The result obtained by the proposed segmentation strategy is evaluated on the 2002 Rich Transcription (RT-02) Evaluation dataset, and a miss rate of 19.47% and a false alarm rate of 16.94% is achieved at the optimal threshold.

Eigenvoice modeling for cross likelihood ratio based speaker clustering : a Bayesian approach

This paper proposes the use of Bayesian approaches with the cross likelihood ratio (CLR) as a criterion for speaker clustering within a speaker diarization system, using eigenvoice modeling techniques. The CLR has previously been shown to be an effective decision criterion for speaker clustering using Gaussian mixture models. Recently, eigenvoice modeling has become an increasingly popular technique, due to its ability to adequately represent a speaker based on sparse training data, as well as to provide an improved capture of differences in speaker characteristics. The integration of eigenvoice modeling into the CLR framework to capitalize on the advantage of both techniques has also been shown to be beneficial for the speaker clustering task. Building on that success, this paper proposes the use of Bayesian methods to compute the conditional probabilities in computing the CLR, thus effectively combining the eigenvoice-CLR framework with the advantages of a Bayesian approach to the diarization problem. Results obtained on the 2002 Rich Transcription (RT-02) Evaluation dataset show an improved clustering performance, resulting in a 33.5% relative improvement in the overall Diarization Error Rate (DER) compared to the baseline system.

Likelihood‐free inference for transmission rates of nosocomial pathogens

This chapter contains sections titled: Introduction Case study: Estimating transmission rates of nosocomial pathogens Models and methods Data analysis and results Discussion References

Bayesian computation via empirical likelihood

Approximate Bayesian computation has become an essential tool for the analysis of complex stochastic models when the likelihood function is numerically unavailable. However, the well-established statistical method of empirical likelihood provides another route to such settings that bypasses simulations from the model and the choices of the approximate Bayesian computation parameters (summary statistics, distance, tolerance), while being convergent in the number of observations. Furthermore, bypassing model simulations may lead to significant time savings in complex models, for instance those found in population genetics. The Bayesian computation with empirical likelihood algorithm we develop in this paper also provides an evaluation of its own performance through an associated effective sample size. The method is illustrated using several examples, including estimation of standard distributions, time series, and population genetics models.

A quasi-maximum likelihood method for estimating the parameters of multivariate diffusions

A quasi-maximum likelihood procedure for estimating the parameters of multi-dimensional diffusions is developed in which the transitional density is a multivariate Gaussian density with first and second moments approximating the true moments of the unknown density. For affine drift and diffusion functions, the moments are exactly those of the true transitional density and for nonlinear drift and diffusion functions the approximation is extremely good and is as effective as alternative methods based on likelihood approximations. The estimation procedure generalises to models with latent factors. A conditioning procedure is developed that allows parameter estimation in the absence of proxies.

Self-reported likelihood of speeding : the effects of attitudes, personality, and perceived legitimacy of enforcement

The positive relationship between speed and crash risk and severity is robust and well-established. While excessive speeding is typically regarded by the public as a common contributing factor in road crashes, speeding remains a common traffic infringement and an arguably socially acceptable behaviour, particularly at low levels over the speed limit. This suggests that other factors potentially contribute to this disparity between crash perceptions and actual behaviours. Previous work has described associations between perceptions of the legitimacy of speed enforcement, attitudes, and how they relate to the likelihood of speeding. This study sought to more closely examine the nature of the relationships between these variables. In total, 293 Queensland drivers participated in a study that examined how demographics, personality variables, attitudes, and perceptions of the legitimacy of enforcement contributed to drivers’ self-reported likelihood of speeding. Results suggested that positive attitudes towards speeding had the greatest impact on likelihood of speeding behaviours. Being younger and higher levels of the personality trait of extraversion were also associated with greater levels of self-reported likelihood of speeding. Attitudes were found to mediate the relationship between perceived legitimacy of speed enforcement and self-reported likelihood of speeding. A subgroup analysis of participants with positive and negative attitudes towards speeding revealed that a differential set of variables were predictive of self-reported likelihood of speeding for the two subgroups. This highlights the potential importance of attitudes in understanding the influence of perceptions of legitimacy of speed enforcement on speeding behaviour, and the need for targeted rather than a ‘one size fits all’ approach to changing attitudes and ultimately behaviour. The findings of the current study help to further understand why some drivers continue to speed.

Shrinkage empirical likelihood estimator in longitudinal analysis with time-dependent covariates-application to modeling the health of Filipino children

The method of generalized estimating equations (GEE) is a popular tool for analysing longitudinal (panel) data. Often, the covariates collected are time-dependent in nature, for example, age, relapse status, monthly income. When using GEE to analyse longitudinal data with time-dependent covariates, crucial assumptions about the covariates are necessary for valid inferences to be drawn. When those assumptions do not hold or cannot be verified, Pepe and Anderson (1994, Communications in Statistics, Simulations and Computation 23, 939–951) advocated using an independence working correlation assumption in the GEE model as a robust approach. However, using GEE with the independence correlation assumption may lead to significant efficiency loss (Fitzmaurice, 1995, Biometrics 51, 309–317). In this article, we propose a method that extracts additional information from the estimating equations that are excluded by the independence assumption. The method always includes the estimating equations under the independence assumption and the contribution from the remaining estimating equations is weighted according to the likelihood of each equation being a consistent estimating equation and the information it carries. We apply the method to a longitudinal study of the health of a group of Filipino children.

Predicting the likelihood of QA failure using treatment plan accuracy metrics

This study used automated data processing techniques to calculate a set of novel treatment plan accuracy metrics, and investigate their usefulness as predictors of quality assurance (QA) success and failure. 151 beams from 23 prostate and cranial IMRT treatment plans were used in this study. These plans had been evaluated before treatment using measurements with a diode array system. The TADA software suite was adapted to allow automatic batch calculation of several proposed plan accuracy metrics, including mean field area, small-aperture, off-axis and closed-leaf factors. All of these results were compared the gamma pass rates from the QA measurements and correlations were investigated. The mean field area factor provided a threshold field size (5 cm2, equivalent to a 2.2 x 2.2 cm2 square field), below which all beams failed the QA tests. The small aperture score provided a useful predictor of plan failure, when averaged over all beams, despite being weakly correlated with gamma pass rates for individual beams. By contrast, the closed leaf and off-axis factors provided information about the geometric arrangement of the beam segments but were not useful for distinguishing between plans that passed and failed QA. This study has provided some simple tests for plan accuracy, which may help minimise time spent on QA assessments of treatments that are unlikely to pass.

Recursive pathways to marginal likelihood estimation with prior-sensitivity analysis

We investigate the utility to computational Bayesian analyses of a particular family of recursive marginal likelihood estimators characterized by the (equivalent) algorithms known as "biased sampling" or "reverse logistic regression" in the statistics literature and "the density of states" in physics. Through a pair of numerical examples (including mixture modeling of the well-known galaxy dataset) we highlight the remarkable diversity of sampling schemes amenable to such recursive normalization, as well as the notable efficiency of the resulting pseudo-mixture distributions for gauging prior-sensitivity in the Bayesian model selection context. Our key theoretical contributions are to introduce a novel heuristic ("thermodynamic integration via importance sampling") for qualifying the role of the bridging sequence in this procedure, and to reveal various connections between these recursive estimators and the nested sampling technique.

What aspects of demographic, personality, attitudes, and perceptions of law enforcement influence self-reported likelihood of drink driving?

The aim of the current study was to examine the associations between a number of individual factors (demographic factors (age and gender), personality factors, risk-taking propensity, attitudes towards drink driving, and perceived legitimacy of drink driving enforcement) and how they influence the self-reported likelihood of drink driving. The second aim of this study was to examine the potential of attitudes mediating the relationship between risk-taking and self-reported likelihood of drink driving. In total, 293 Queensland drivers volunteered to participate in an online survey that assessed their self-reported likelihood to drink drive in the next month, demographics, traffic-related demographics, personality factors, risk-taking propensity, attitudes towards drink driving, and perceived legitimacy of drink driving enforcement. An ordered logistic regression analysis was utilised to evaluate the first aim of the study; at the first step the demographic variables were entered; at step two the personality and risk-taking were entered; at the third step, the attitudes and perceptions of legitimacy variables were entered. Being a younger driver and having a high risk-taking propensity were related to self-reported likelihood of drink driving. However, when the attitudes variable was entered, these individual factors were no longer significant; with attitudes being the most important predictor of self-reported drink driving likelihood. A significant mediation model was found with the second aim of the study, such that attitudes mediated the relationship between risk-taking and self-reported likelihood of drink driving. Considerable effort and resources are utilised by traffic authorities to reducing drink driving on the Australian road network. Notwithstanding these efforts, some participants still had some positive attitudes towards drink driving and reported that they were likely to drink drive in the future. These findings suggest that more work is needed to address attitudes regarding the dangerousness of drink driving.

Novel use of faecal sterols to assess human faecal contamination in Antarctica : a likelihood assessment matrix for environmental monitoring

Wastewater containing human sewage is often discharged with little or no treatment into the Antarctic marine environment. Faecal sterols (primarily coprostanol) in sediments have been used for assessment of human sewage contamination in this environment, but in situ production and indigenous faunal inputs can confound such determinations. Using gas chromatography with mass spectral detection profiles of both C27 and C29 sterols, potential sources of faecal sterols were examined in nearshore marine sediments, encompassing sites proximal and distal to the wastewater outfall at Davis Station. Faeces from indigenous seals and penguins were also examined. Faeces from several indigenous species contained significant quantities of coprostanol but not 24-ethylcoprostanol, which is present in human faeces. In situ coprostanol and 24-ethylcoprostanol production was identified by co-production of their respective epi isomers at sites remote from the wastewat er source and in high total organic matter sediments. A C 29 sterols-based polyphasic likelihood assessment matrix for human sewage contamination is presented, which distinguishes human from local fauna faecal inputs and in situ production in the Antarctic environment. Sewage contamination was detected up to 1.5 km from Davis Station.

Predicting the likelihood of non-healing : a venous leg ulcer risk assessment tool

This project developed, validated and tested reliability of a risk assessment tool to predict the risk of failure to heal of patients with venous leg ulcers within 24 weeks. The risk assessment tool will allow clinicians to be able to determine realistic outcomes for their patients, promote early healing and potentially avoid weeks of inappropriate therapy. The tool will also assist in addressing specific risk factors and guide decisions on early, alternative, tailored interventions.

The elaboration likelihood model : review, critique and research agenda

Purpose: The purpose of this paper is to review, critique and develop a research agenda for the Elaboration Likelihood Model (ELM). The model was introduced by Petty and Cacioppo over three decades ago and has been modified, revised and extended. Given modern communication contexts, it is appropriate to question the model’s validity and relevance. Design/methodology/approach: The authors develop a conceptual approach, based on a fully comprehensive and extensive review and critique of ELM and its development since its inception. Findings: This paper focuses on major issues concerning the ELM. These include model assumptions and its descriptive nature; continuum questions, multi-channel processing and mediating variables before turning to the need to replicate the ELM and to offer recommendations for its future development. Research limitations/implications: This paper offers a series of questions in terms of research implications. These include whether ELM could or should be replicated, its extension, a greater conceptualization of argument quality, an explanation of movement along the continuum and between central and peripheral routes to persuasion, or to use new methodologies and technologies to help better understanding consume thinking and behaviour? All these relate to the current need to explore the relevance of ELM in a more modern context. Practical implications: It is time to question the validity and relevance of the ELM. The diversity of on- and off-line media options and the variants of consumer choice raise significant issues. Originality/value: While the ELM model continues to be widely cited and taught as one of the major cornerstones of persuasion, questions are raised concerning its relevance and validity in 21st century communication contexts.

A Likelihood-Maximizing Framework for Enhanced In-Car Speech Recognition Based on Speech Dialog System Interaction

Speech recognition in car environments has been identified as a valuable means for reducing driver distraction when operating noncritical in-car systems. Under such conditions, however, speech recognition accuracy degrades significantly, and techniques such as speech enhancement are required to improve these accuracies. Likelihood-maximizing (LIMA) frameworks optimize speech enhancement algorithms based on recognized state sequences rather than traditional signal-level criteria such as maximizing signal-to-noise ratio. LIMA frameworks typically require calibration utterances to generate optimized enhancement parameters that are used for all subsequent utterances. Under such a scheme, suboptimal recognition performance occurs in noise conditions that are significantly different from that present during the calibration session – a serious problem in rapidly changing noise environments out on the open road. In this chapter, we propose a dialog-based design that allows regular optimization iterations in order to track the ever-changing noise conditions. Experiments using Mel-filterbank noise subtraction (MFNS) are performed to determine the optimization requirements for vehicular environments and show that minimal optimization is required to improve speech recognition, avoid over-optimization, and ultimately assist with semireal-time operation. It is also shown that the proposed design is able to provide improved recognition performance over frameworks incorporating a calibration session only.