We present a probabilistic model of a user's search history and a target query reformulation. We derive a simple transitive similarity algorithm for disambiguating queries and improving history-based query reformulation accuracy. We compare the merits of this approach to other methods and present results on both examples assessed by human editors and on automatically-labeled click data.

We present a generalization of dynamic Bayesian networks to concisely describe complex probability distributions such as in problems with multiple interacting variable-length streams of random variables. Our framewor k incorporates recent graphical model constructs to account for existence uncert ainty, value-specific independence, aggregation relationships, and local and global constraints, while still retaining a Bayesian network interpretation and effic ient inference and learning techniques. We introduce one such general technique, which is an extension of Value Elimination, a backtracking search inference algo rithm. Multi-dynamic Bayesian networks are motivated by our work on Statistical Machine Translation (MT). We present results on MT word alignment in support of our claim that MDBNs are a promising framework for the rapid prototyping of new MT systems.

Sitting at the intersection between statistics and machine learning, Dynamic Bayesian Networks have been applied with much success in many domains, such as speech recognition, vision, and computational biology. While Natural Language Processing increasingly relies on statistical methods, we think they have yet to use Graphical Models to their full potential. In this paper, we report on experiments in learning edit distance costs using Dynamic Bayesian Networks and present results on a pronunciation classification task. By exploiting the ability within the DBN framework to rapidly explore a large model space, we obtain a 40% reduction in error rate compared to a previous transducer-based method of learning edit distance.

The modulation spectrum is a promising method to incorporate dynamic information in pattern classification. It contains important cues about the nonstationary content of a signal and yields complementary improvements when it is combined with conventional features derived from short-term analysis. Many prior modulation spectrum approaches are based on uniform modulation frequency decomposition. The drawbacks of these approaches are high dimensionality and a lack of a connection to human perception of modulation. The paper presents multi-scale modulation frequency decomposition and shows an improvement over the standard modulation spectrum in a digital communication signal classification task. Features derived from this representation provide lower classification error rates than those from a constant-bandwidth modulation spectrum, whether used alone or in combination with short-term features.

In recent years there has been growing interest in discriminative parameter training techniques, resulting from notable improvements in speech recognition performance on tasks ranging in size from digit recognition to Switchboard. Typified by Maximum Mutual Information training, these methods assume a fixed statistical modeling structure, and then optimize only the associated numerical parameters (such as means, variances, and transition matrices). In this paper, we explore the significantly different methodology of discriminative structure learning. Here, the fundamental dependency relationships between random variables in a probabilistic model are learned in a discriminative fashion, and are learned separately from the numerical parameters. Tn order to apply the principles of structural discriminability, we adopt the framework of graphical models, which allows an arbitrary set of variables with arbitrary conditional independence relationships to be modeled at each time frame. We present results using a new graphical modeling toolkit (described in a companion paper) from the recent 2001 Johns Hopkins Summer Workshop. These results indicate that significant gains result from discriminative structural analysis of both conventional MFCC and novel AM-FM features on the Aurora continuous digits task.

It is important to produce automatic speech recognition (ASR) systems that use as few computational and memory resources as possible, especially in low-memory/low-power environments such as for personal digital assistants. One way to achieve this is through parameter quantization. In this work, we compare a variety of novel subvector clustering procedures for ASR system parameter quantization. Specifically, we look at systematic data-driven subvector selection techniques based on entropy minimization, and compare performance on a 150-word isolated word speech recognition task. While the optimal entropy-minimizing quantization methods are intractable, we show that although several of our heuristic techniques are elaborate in their attempt to approximate the optimal clustering, a simple scalar quantization scheme using separate codebooks performs remarkably well.