Marc Sobel

A Test for Measuring Gustatory Function

Objectives/Hypothesis: The purpose of this study was to determine the usefulness of edible taste strips for measuring human gustatory function.

Using the Particle Filter Approach to Building Partial Correspondences Between Shapes

Constructing correspondences between points characterizing one shape with those characterizing another is crucial to understanding what the two shapes have in common. These correspondences are the basis for most alignment processes and shape similarity measures. In this paper we use particle filters to establish perceptually correct correspondences between point sets characterizing shapes. Local shape feature descriptors are used to establish the probability that a point on one shape corresponds to a point on the other shape. Global correspondence structures are calculated using additional constraints on domain knowledge. Domain knowledge is characterized by prior distributions which serve to characterize hypotheses about the global relationships between shapes. These hypotheses are formulated online. This means global constraints are learnt during the particle filtering process, which makes the approach especially interesting for applications where global constraints are hard to define a priori. As an example for such a case, experiments demonstrate the performance of our approach on partial shape matching.

Merging maps of multiple robots

Merging local maps, acquired by multiple robots, into a global map, (also known as map merging) is one of the important issues faced by virtually all cooperative exploration techniques. We present a novel and simple solution to the problem of map merging by reducing it to the problem of SLAM of a single ¿virtual¿ robot. The individual local maps and their shape information constitute the sensor information for the virtual robot. This approach allows us to adapt the framework of Rao-Blackwellized particle filtering used in SLAM of a single robot for the problem of map merging.

New EM derived from Kullback-Leibler divergence

We introduce a new EM framework in which it is possible not only to optimize the model parameters but also the number of model components. A key feature of our approach is that we use nonparametric density estimation to improve parametric density estimation in the EM framework. While the classical EM algorithm estimates model parameters empirically using the data points themselves, we estimate them using nonparametric density estimates.There exist many possible applications that require optimal adjustment of model components. We present experimental results in two domains. One is polygonal approximation of laser range data, which is an active research topic in robot navigation. The other is grouping of edge pixels to contour boundaries, which still belongs to unsolved problems in computer vision.

Correspondences between parts of shapes with particle filters

Given two shapes, the correspondence between distinct visual features is the basis for most alignment processes and shape similarity measures. This paper presents an approach introducing particle filters to establish perceptually correct correspondences between point sets representing shapes. Local shape feature descriptors are used to establish correspondence probabilities. The global correspondence structure is calculated using additional constraints based on domain knowledge. Domain knowledge is characterized as prior distributions expressing hypotheses about the global relationships between shapes. These hypotheses are generated during the iterative particle filtering process. Experiments using standard alignment techniques, based on the given correspondence relationships, demonstrate the advantages of this approach.

Complete ranking procedures with appropriate loss functions

This paper treats the problem of comparing different evaluations of procedures which rank the variances of k normal populations. Procedures are evaluated on the basis of appropriate loss functions for a particular goal. The goal considered involves ranking the variances of k independent normal populations when the corresponding population means are unknown. The variances are ranked by selecting samples of size n from each population and using the sample variances to obtain the ranking. Our results extend those of various authors who looked at the narrower problem of evaluating the standard proceduv 3 associated with selecting the smallest of the population variances (see e.g.,P. Somerville (1975)). Different loss functions (both parametric and non-parametric) appropriate to the particular goal under consideration are proposed. Procedures are evaluated by the performance of their risk over a particular preference zone. The sample size n, the least favorable parametric configuration, and the maximum value o...

A left-to-right HDP-HMM with HDPM emissions

Nonparametric Bayesian models use a Bayesian framework to learn the model complexity automatically from the data and eliminate the need for a complex model selection process. The Hierarchical Dirichlet Process hidden Markov model (HDP-HMM) is the nonparametric Bayesian equivalent of an HMM. However, HDP-HMM is restricted to an ergodic topology and uses a Dirichlet Process Model (DPM) to achieve a mixture distribution-like model. For applications such as speech recognition, where we deal with ordered sequences, it is desirable to impose a left-to-right structure on the model to improve its ability to model the sequential nature of the speech signal. In this paper, we introduce three enhancements to HDP-HMM: (1) a left-to-right structure: needed for sequential decoding of speech, (2) non-emitting initial and final states: required for modeling finite length sequences, (3) HDP mixture emissions: allows sharing of data across states. The latter is particularly important for speech recognition because Gaussian mixture models have been very effective at modeling speaker variability. Further, due to the nature of language, some models occur infrequently and have a small number of data points associated with them, even for large corpora. Sharing allows these models to be estimated more accurately. We demonstrate that this new HDP-HMM model produces a 15% increase in likelihoods and a 15% relative reduction in error rate on a phoneme classification task based on the TIMIT Corpus.

Applications of Dirichlet Process Mixtures to speaker adaptation

Balancing unique acoustic characteristics of a speaker such as identity and accent, with general acoustic behavior that describes phoneme identity, is one of the great challenges in applying nonparametric Bayesian approaches to speaker adaptation. The Dirichlet Process Mixture (DPM) is a relatively new model that provides an elegant framework in which individual characteristics can be balanced with aggregate behavior without diluting the quality of the individual models. Unlike Gaussian Mixture models (GMMs), which tend to smear multimodal behavior through averaging, the DPM model attempts to preserve unique behaviors through use of an infinite mixture model. In this paper, we present some exploratory research on applying these models to the acoustic modeling component of the speaker adaptation problem. DPM based models are shown to provide up to 10% reduction in WER over maximum likelihood linear regression (MLLR) on a speaker adaptation task based on the Resource Management database.

Bayesian auxiliary particle filters for estimating neural tuning parameters

A common challenge in neural engineering is to track the dynamic parameters of neural tuning functions. This work introduces the application of Bayesian auxiliary particle filters for this purpose. Based on Monte-Carlo filtering, Bayesian auxiliary particle filters use adaptive methods to model the prior densities of the state parameters being tracked. The observations used are the neural firing times, modeled here as a Poisson process, and the biological driving signal. The Bayesian auxiliary particle filter was evaluated by simultaneously tracking the three parameters of a hippocampal place cell and compared to a stochastic state point process filter. It is shown that Bayesian auxiliary particle filters are substantially more accurate and robust than alternative methods of state parameter estimation. The effects of time-averaging on parameter estimation are also evaluated.

Correspondences of point sets using Particle Filters

The paper shows how particle filters can be used to establish visually consistent partial correspondences between similar features in unrestricted 2D point sets representing shapes. Given an update rule, the PF system has the advantage that global constraints can be learned. We motivate and define the update rule for the given task and show its superior performance in comparison to a globally unrestricted approach.