Maximum entropy direct models for speech recognition

Traditional statistical models for speech recognition have mostly been based on a Bayesian framework using generative models such as hidden Markov models (HMMs). This paper focuses on a new framework for speech recognition using maximum entropy direct modeling, where the probability of a state or word sequence given an observation sequence is computed directly from the model. In contrast to HMMs, features can be asynchronous and overlapping. This model therefore allows for the potential combination of many different types of features, which need not be statistically independent of each other. In this paper, a specific kind of direct model, the maximum entropy Markov model (MEMM), is studied. Even with conventional acoustic features, the approach already shows promising results for phone level decoding. The MEMM significantly outperforms traditional HMMs in word error rate when used as stand-alone acoustic models. Preliminary results combining the MEMM scores with HMM and language model scores show modest improvements over the best HMM speech recognizer.     

OWL-Eu: Adding customised datatypes into OWL

Although OWL is rather expressive, it has a very serious limitation on datatypes; i.e., it does not support customised datatypes. It has been pointed out that many potential users will not adopt OWL unless this limitation is overcome, and the W3C Semantic Web Best Practices and Development Working Group has set up a task force to address this issue. This paper makes the following two contributions: (i) it provides a brief summary of OWL-related datatype formalisms, and (ii) it provides a decidable extension of OWL DL, called OWL-Eu, that supports customised datatypes. A detailed proof of the decidability of OWL-Eu is presented.     

Customization of a Commercially Available Prep Scale SFC System to Provide Enhanced Capabilities

Preparative Scale Supercritical Fluid Chromatography is emerging as a powerful alternative to HPLC for the purification and separation of complex chemical reaction mixtures. Advantages include greatly reduced solvent usage (and thus lower cost and environmental impact), higher throughput, and in some cases higher resolution. While there are commercially available prep SFC instruments, none currently offer all the features desired by many medicinal chemists engaged in the drug discovery process. These include: the ability to collect an unlimited number of fractions per sample with high recovery and negligible carryover, fully automated capacity to collect several hundred fractions, and the ability to collect fractions into the same disposable test tubes and racks which are already employed in HPLC. This article describes the customization of a preparatory scale SFC system purchased from Berger Instruments, Inc., Newark, DE. (a subsidiary Mettler-Toledo International, Inc., of Greifensee, Switzerland) in order to provide these capabilities.     

Low-Power Parallel Video Compression Architecture for a Single-Chip Digital CMOS Camera

A low-power, large-scale parallel video compression architecture for a single-chip digital CMOS camera is discussed in this paper. This architecture is designed for highly computationally intensive image and video processing tasks necessary to support video compression. Two designs of this architecture, an MPEG2 encoder and a DV encoder, are presented. At an image resolution of 640 × 480 pixels (MPEG2) and 720 × 576 (DV) and a frame rate of 25 to 30 frames per second, a computational throughput of up to 1.8 billion operations per second (BOPS) is required. This is supported in the proposed architecture using a 40 MHz clock and an array of 40 to 45 parallel processors implemented in a 0.2 m CMOS technology and with a 1.5 V supply voltage. Power consumption is significantly reduced through the single-chip integration of the CMOS photo sensors, the embedded DRAM technology, and the proposed pipelined parallel processors. The parallel processors consume approximately 45 mW of power resulting a power efficiency of 40 BOPS/W.     

Metabolite-sensitive holographic biosensors

A new type of biosensor that combines the inexpensiveness and mass-produceability of reflection holograms with the selectivity and specificity of enzymes is described. pH-sensitive holographic sensors were fabricated from ionizable monomers incorporated into thin, polymeric, hydrogel films which were transformed into volume holograms using a diffusion method coupled with holographic recording, using a frequency-doubled Nd:YAG laser (532 nm). These holograms were used as transducer systems to monitor the pH changes associated with specific enzymatic reactions to construct prototype urea- and penicillin-sensitive biosensors. The diffraction wavelength (color) of the holographic biosensors was used to characterize their shrinkage and swelling behavior as a function of analyte concentration. The potential of these sensors for the measurement of the clinically and industrially important metabolites urea and penicillin G is demonstrated.     

Glucose-sensitive holographic sensors for monitoring bacterial growth

A glucose sensor comprising a reflection hologram incorporated into a thin, acrylamide hydrogel film bearing the cis-diol binding ligand, 3-acrylamidophenylboronic acid (3-APB), is described. The diffraction wavelength (color) of the hologram changes as the polymer swells upon binding cis-diols. The effect of various concentrations of glucose, a variety of mono- and disaccharides, and the alpha-hydroxy acid, lactate, on the holographic response was investigated. The sensor displayed reversible changes in diffraction wavelength as a function of cis-diol concentration, with the sensitivity of the system being dependent on the cis-diol tested. The effect of varying 3-APB concentration in the hydrogel on the holographic response to glucose was investigated, and maximum sensitivity was observed at a functional monomer concentration of 20 mol %. The potential for using this holographic sensor to detect real-time changes in bacterial cell metabolism was demonstrated by monitoring the germination and subsequent vegetative growth of Bacillus subtilis spores.     

Simultaneous registration and activation detection for fMRI

Registration using the least-squares cost function is sensitive to the intensity fluctuations caused by the blood oxygen level dependent (BOLD) signal in functional MRI (fMRI) experiments, resulting in stimulus-correlated motion errors. These errors are severe enough to cause false-positive clusters in the activation maps of datasets acquired from 3T scanners. This paper presents a new approach to resolving the coupling between registration and activation. Instead of treating the two problems as individual steps in a sequence, they are combined into a single least-squares problem and are solved simultaneously. Robustness tests on a variety of simulated three-dimensional EPI datasets show that the stimulus-correlated motion errors are removed, resulting in a substantial decrease in false-positive and false-negative activation rates. The new method is also shown to decorrelate the motion estimates from the stimulus by testing it on different in vivo fMRI datasets acquired from two different 3T scanners.     

Verifying feature models using OWL

Feature models are widely used in domain engineering to capture common and variant features among systems in a particular domain. However, the lack of a formal semantics and reasoning support of feature models has hindered the development of this area. Industrial experiences also show that methods and tools that can support feature model analysis are badly appreciated. Such reasoning tool should be fully automated and efficient. At the same time, the reasoning tool should scale up well since it may need to handle hundreds or even thousands of features a that modern software systems may have. This paper presents an approach to modeling and verifying feature diagrams using Semantic Web OWL ontologies. We use OWL DL ontologies to precisely capture the inter-relationships among the features in a feature diagram. OWL reasoning engines such as FaCT++ are deployed to check for the inconsistencies of feature configurations fully automatically. Furthermore, a general OWL debugger has been developed to tackle the disadvantage of lacking debugging aids for the current OWL reasoner and to complement our verification approach. We also developed a CASE tool to facilitate visual development, interchange and reasoning of feature diagrams in the Semantic Web environment.     

MVA Processing of Speech Features

In this paper, we investigate a technique consisting of mean subtraction, variance normalization and time sequence filtering. Unlike other techniques, it applies auto-regression moving-average (ARMA) filtering directly in the cepstral domain. We call this technique mean subtraction, variance normalization, and ARMA filtering (MVA) post-processing, and speech features with MVA post-processing are called MVA features. Overall, compared to raw features without post-processing, MVA features achieve an error rate reduction of 45% on matched tasks and 65% on mismatched tasks on the Aurora 2.0 noisy speech database, and an average 57% error reduction on the Aurora 3.0 database. These improvements are comparable to the results of much more complicated techniques even though MVA is relatively simple and requires practically no additional computational cost. In this paper, in addition to describing MVA processing, we also present a novel analysis of the distortion of mel-frequency cepstral coefficients and the log energy in the presence of different types of noise. The effectiveness of MVA is extensively investigated with respect to several variations: the configurations used to extract and the type of raw features, the domains where MVA is applied, the filters that are used, the ARMA filter orders, and the causality of the normalization process. Specifically, it is argued and demonstrated that MVA works better when applied to the zeroth-order cepstral coefficient than to log energy, that MVA works better in the cepstral domain, that an ARMA filter is better than either a designed finite impulse response filter or a data-driven filter, and that a five-tap ARMA filter is sufficient to achieve good performance in a variety of settings. We also investigate and evaluate a multi-domain MVA generalization     

Normality evaluations and their relation to personality traits and well-being.

Despite frequent discussions of what it means to be normal in clinical, social, and personality psychology theory, the characteristics of individuals who call themselves normal are little understood. In 5 studies, the authors investigated various hypotheses concerning the nature of normality evaluations. The authors add to recent evidence that normality evaluations represent a distinct dimension of evaluative judgments, showing self-judgments of being normal (versus strange) to be relatively independent from self-judgments of being average (versus unique). Normality evaluations showed positive relations with communal traits such as agreeableness, conscientiousness, and emotional stability, and were negatively related to openness to experience. Supporting a broader hypothesis that normality evaluations may be involved in directing or motivating personality development processes, normality evaluations were positively associated with well-being and a sense of fitting in with one's peers, and individuals who felt abnormal felt a heightened sense that they needed to improve their personality. Finally, the personality correlates of normality evaluations were found to change over the lifespan, largely in parallel with the actual mean-level development of personality traits with age. (PsycINFO Database Record (c) 2010 APA, all rights reserved)