Alan Wisler

Empirically Estimable Classification Bounds Based on a Nonparametric Divergence Measure

Information divergence functions play a critical role in statistics and information theory. In this paper we show that a nonparametric f-divergence measure can be used to provide improved bounds on the minimum binary classification probability of error for the case when the training and test data are drawn from the same distribution and for the case where there exists some mismatch between training and test distributions. We confirm these theoretical results by designing feature selection algorithms using the criteria from these bounds and by evaluating the algorithms on a series of pathological speech classification tasks.

Domain invariant speech features using a new divergence measure

Existing speech classification algorithms often perform well when evaluated on training and test data drawn from the same distribution. In practice, however, these distributions are not always the same. In these circumstances, the performance of trained models will likely decrease. In this paper, we discuss an underutilized divergence measure and derive an estimable upper bound on the test error rate that depends on the error rate on the training data and the distance between training and test distributions. Using this bound as motivation, we develop a feature learning algorithm that aims to identify invariant speech features that generalize well to data similar to, but different from, the training set. Comparative results confirm the efficacy of the algorithm on a set of cross-domain speech classification tasks.

The relationship between perceptual disturbances in dysarthric speech and automatic speech recognition performance

State-of-the-art automatic speech recognition (ASR) engines perform well on healthy speech; however recent studies show that their performance on dysarthric speech is highly variable. This is because of the acoustic variability associated with the different dysarthria subtypes. This paper aims to develop a better understanding of how perceptual disturbances in dysarthric speech relate to ASR performance. Accurate ratings of a representative set of 32 dysarthric speakers along different perceptual dimensions are obtained and the performance of a representative ASR algorithm on the same set of speakers is analyzed. This work explores the relationship between these ratings and ASR performance and reveals that ASR performance can be predicted from perceptual disturbances in dysarthric speech with articulatory precision contributing the most to the prediction followed by prosody.

Removing data with noisy responses in regression analysis

In regression analysis, outliers in the data can induce a bias in the learned function, resulting in larger errors. In this paper we derive an empirically estimable bound on the regression error based on a Euclidean minimum spanning tree generated from the data. Using this bound as motivation, we propose an iterative approach to remove data with noisy responses from the training set. We evaluate the performance of the algorithm on experiments with real-world pathological speech (speech from individuals with neurogenic disorders). Comparative results show that removing noisy examples during training using the proposed approach yields better predictive performance on out-of- sample data.

Direct Estimation of Density Functionals Using a Polynomial Basis

A number of fundamental quantities in statistical signal processing and information theory can be expressed as integral functions of two probability density functions. Such quantities are called density functionals as they map density functions onto the real line. For example, information divergence functions measure the dissimilarity between two probability density functions and are useful in a number of applications. Typically, estimating these quantities requires complete knowledge of the underlying distribution followed by multidimensional integration. Existing methods make parametric assumptions about the data distribution or use nonparametric density estimation followed by high-dimensional integration. In this paper, we propose a new alternative. We introduce the concept of “data-driven basis functions”—functions of distributions whose value we can estimate given only samples from the underlying distributions without requiring distribution fitting or direct integration. We derive a new data-driven complete basis that is similar to the deterministic Bernstein polynomial basis and develop two methods for performing basis expansions of functionals of two distributions. We also show that the new basis set allows us to approximate functions of distributions as closely as desired. Finally, we evaluate the methodology by developing data-driven estimators for the Kullback–Leibler divergences and the Hellinger distance and by constructing empirical estimates of tight bounds on the Bayes error rate.

Empirically-estimable multi-class classification bounds

In this paper, we extend previously developed non-parametric bounds on the Bayes risk in binary classification problems to multi-class problems. In comparison with the well-known Bhattacharyya bound which is typically calculated by employing parametric assumptions, the bounds proposed in this paper are directly estimable from data, provably tighter, and more robust to different types of data. We verify the tightness and validity of this bound using an illustrative synthetic example, and further demonstrate its value by incorporating it into a feature selection algorithm which we apply to the real-world problem of distinguishing between different neuro-motor disorders based on sentence-level speech data.

A Framework for Automated Marmoset Vocalization Detection And Classification

This paper describes a novel framework for automated marmoset vocalization detection and classification from within long audio streams recorded in a noisy animal room, where multiple marmosets are housed. To overcome the challenge of limited manually annotated data, we implemented a data augmentation method using only a small number of labeled vocalizations. The feature sets chosen have the desirable property of capturing characteristics of the signals that are useful in both identifying and distinguishing marmoset vocalizations. Unlike many previous methods, feature extraction, call detection, and call classification in our system are completely automated. The system maintains a good performance of 20% equal error detection rate using data with high number of noise events and 15% of classification error. Performance can be further improved with additional labeled training data. Because this extensible system is capable of identifying both positive and negative welfare indicators, it provides a powerful framework for non-human primate welfare monitoring as well as behavior assessment.

Noise robust dysarthric speech classification using domain adaptation

This paper will investigate viability of a screening application that could be used to identify individuals with Dysarthria from among a larger population using sentence-level speech data. This task presents a number of challenged particularly if we aim to identify the disorder in the earlier stages before the more significant symptoms have begun to manifest themselves. A principal challenge in this task is acheiving robustness to the large number of confounding variables such as gender, age, accent, speaking style, and recording conditions. All of these variables will affect an individuals speech in a manner unrelated to the disorder, and identifying what information is relevant to the disorder amongst these confounding variables given the limited amount of data that is available in this regime presents a major engineering challenge. In this paper we will focus on achieving robustness to different types and levels of noise by employing a feature selection algorithm that attempts to minimize a non-parametric upper bound of the error in the noisy condition. This is a crucial problem to solve as the clean recording conditions used in data collection are typically a poor reflection of the type of data that will be encountered upon deployment.