Extending boosting for large scale spoken language understanding

We propose three methods for extending the Boosting family of classifiers motivated by the real-life problems we have encountered. First, we propose a semisupervised learning method for exploiting the unlabeled data in Boosting. We then present a novel classification model adaptation method. The goal of adaptation is optimizing an existing model for a new target application, which is similar to the previous one but may have different classes or class distributions. Finally, we present an efficient and effective cost-sensitive classification method that extends Boosting to allow for weighted classes. We evaluated these methods for call classification in the AT&;T VoiceTone^® spoken language understanding system. Our results indicate that it is possible to obtain the same classification performance by using 30% less labeled data when the unlabeled data is utilized through semisupervised learning. Using model adaptation we can achieve the same classification accuracy using less than half of the labeled data from the new application. Finally, we present significant improvements in the “important” (i.e., higher weighted) classes without a significant loss in overall performance using the proposed cost-sensitive classification method.     

Polarization Mode Dispersion of Installed Fibers

Polarization mode dispersion (PMD), a potentially limiting impairment in high-speed long-distance fiber-optic communication systems, refers to the distortion of propagating optical pulses due to random birefringences in an optical system. Because these perturbations (which can be introduced through manufacturing imperfections, cabling stresses, installation procedures, and environmental sensitivities of fiber and other in-line components) are unknowable and continually changing, PMD is unique among optical impairments. This makes PMD both a fascinating research subject and potentially one of the most challenging technical obstacles for future optoelectronic transmission. Mitigation and compensation techniques, proper emulation, and accurate prediction of PMD-induced outage probabilities critically depend on the understanding and modeling of the statistics of PMD in installed links. Using extensive data on buried fibers used in long-haul high-speed links, the authors discuss the proposition that most of the temporal PMD changes that are observed in installed routes arise primarily from a relatively small number of "hot spots" along the route that are exposed to the ambient environment, whereas the buried shielded sections remain largely stable for month-long time periods. It follows that the temporal variations of the differential group delay for any given channel constitute a distinct statistical distribution with its own channel-specific mean value. The impact of these observations on outage statistics is analyzed, and the implications for future optoelectronic fiber-based transmission are discussed     

Exact 3D boundary representation in finite element analysis based on Cartesian grids independent of the geometry

This paper proposes a novel Immersed Boundary Method where the embedded domain is exactly described by using its Computer‐Aided Design (CAD) boundary representation with Non‐Uniform Rational B‐Splines (NURBS) or T‐splines. The common feature with other immersed methods is that the current approach substantially reduces the burden of mesh generation. In contrast, the exact boundary representation of the embedded domain allows to overcome the major drawback of existing immersed methods that is the inaccurate representation of the physical domain. A novel approach to perform the numerical integration in the region of the cut elements that is internal to the physical domain is presented and its accuracy and performance evaluated using numerical tests. The applicability, performance, and optimal convergence of the proposed methodology is assessed by using numerical examples in three dimensions. It is also shown that the accuracy of the proposed methodology is independent on the CAD technology used to describe the geometry of the embedded domain. Copyright © 2015 John Wiley & Sons, Ltd.     

Design and implementation of an ultrasonic sensor for rapid monitoring of industrial malolactic fermentation of wines

Ultrasound is an emerging technology that can be applied to monitor food processes. However, ultrasonic techniques are usually limited to research activities within a laboratory environment and they are not extensively used in industrial processes. The aim of this paper is to describe a novel ultrasonic sensor designed to monitor physical–chemical changes that occur in wines stored in industrial tanks. Essentially, the sensor consists of an ultrasonic transducer in contact with a buffer rod, mounted inside a stainless steel tube section. This structure allows the ultrasonic sensor to be directly installed in stainless steel tanks of an industrial plant. The operating principle of this design is based on the measurement of ultrasonic velocity of propagation. To test its proper operation, the sensor has been used to measure changes of concentration in aqueous samples and to monitor the progress of a malolactic fermentation of red wines in various commercial wineries. Results show the feasibility of using this sensor for monitoring malolactic fermentations in red wines placed in industrial tanks.     

Optimization of Selenium-Enriched Mycelium of Lentinula edodes (Berk.) Pegler as a Food Supplement

Our goal was to optimize the growth conditions of submerged mycelial cultures of Lentinula edodes (Shiitake mushroom) in order to obtain a new dietary supplement enriched in selenium We designed a process technology in which mycelial cultures were cultivated in media composed of beet molasses, 10%; liquid stillage, 5%; corn steep liquor, 0.15%; and KH₂PO₄, 0.3%, enriched with selenium in concentrations ranging from 0 to 100 μg/mL by the addition of sodium selenite of selenic acid. Se concentrations in mycelial dry mass rose from 0.001 mg/g (mycelia cultivated in media containing no selenium) to 50 mg/g (mycelia cultivated in medium containing 100 μg Se/mL). The highest mycelial specific growth rate (0.46/day) was recorded when the concentration of selenium in the medium was lower than 20 μg/mL. In vitro, estimated Se bioavailabilities from selenized mycelium strongly depended on its preparation in a proper manner and were 60% and 82% for dried mycelium and mycelial lyophilizate, respectively. Our results suggest that these optimized culture conditions could be applied to obtain a new Se-enriched dietary supplement.     

The AT&T spoken language understanding system

Spoken language understanding (SLU) aims at extracting meaning from natural language speech. Over the past decade, a variety of practical goal-oriented spoken dialog systems have been built for limited domains. SLU in these systems ranges from understanding predetermined phrases through fixed grammars, extracting some predefined named entities, extracting users' intents for call classification, to combinations of users' intents and named entities. In this paper, we present the SLU system of VoiceTone/spl reg/ (a service provided by AT&T where AT&T develops, deploys and hosts spoken dialog applications for enterprise customers). The SLU system includes extracting both intents and the named entities from the users' utterances. For intent determination, we use statistical classifiers trained from labeled data, and for named entity extraction we use rule-based fixed grammars. The focus of our work is to exploit data and to use machine learning techniques to create scalable SLU systems which can be quickly deployed for new domains with minimal human intervention. These objectives are achieved by 1) using the predicate-argument representation of semantic content of an utterance; 2) extending statistical classifiers to seamlessly integrate hand crafted classification rules with the rules learned from data; and 3) developing an active learning framework to minimize the human labeling effort for quickly building the classifier models and adapting them to changes. We present an evaluation of this system using two deployed applications of VoiceTone/spl reg/.     

Mobile robot design in education

This paper describes the course on robot design that electronics systems engineers take in their last semester at the Instituto Tecnologico de Estudios Superiores de Monterrey (ITESM) in Mexico. In this course, the use of project oriented learning (POL) and collaborative learning are proposed. The paper also describes the design and implementation of a modular, low-cost, three-wheeled autonomous robotic platform, to serve as a base platform from which different applications, educational and research, could be mounted.     

Learnability and Definability in Trees and Similar Structures

We prove upper bounds for combinatorial parameters of finite relational structures, related to the complexity of learning a definable set. We show that monadic second-order (MSO) formulas with parameters have bounded Vapnik–Chervonenkis dimension over structures of bounded clique-width, and first-order formulas with parameters have bounded Vapnik–Chervonenkis dimension over structures of bounded local clique-width (this includes planar graphs). We also show that MSO formulas of a fixed size have bounded strong consistency dimension over MSO formulas of a fixed larger size, for labeled trees. These bounds imply positive learnability results for the PAC and equivalence query learnability of a definable set over these structures. The proofs are based on bounds for related definability problems for tree automata.     

On the selection of variables for qualitative modelling of dynamical systems

Behavioural modelling of physical systems from observations of their input/output behaviour is an important task in engineering. Such models are needed for fault monitoring as well as intelligent control of these systems. The paper addresses one subtask of behavioural modelling, namely the selection of input variables to be used in predicting the behaviour of an output variable. A technique that is well suited for qualitative behavioural modelling and simulation of physical systems is Fuzzy Inductive Reasoning (FIR), a methodology based on General System Theory. Yet, the FIR modelling methodology is of exponential computational complexity, and therefore, it may be useful to consider other approaches as booster techniques for FIR. Different variable selection algorithms: the method of the unreconstructed variance for the best reconstruction, methods based on regression coefficients (OLS, PCR and PLS) and other methods as Multiple Correlation Coefficients (MCC), Principal Components Analysis (PCA) and Cluster analysis are discussed and compared to each other for use in predicting the behaviour of a steam generator. The different variable selection algorithms previously named are then used as booster techniques for FIR. Some of the used linear techniques have been found to be non-effective in the task of selecting variables in order to compute a posterior FIR model. Methods based on clustering seem particularly well suited for pre-selecting subsets of variables to be used in a FIR modelling and simulation effort.