Maximum likelihood Linear Programming Data Fusion for Speaker Recognition

Biometric system performance can be improved by means of data fusion. Several kinds of information can be fused in order to obtain a more accurate classification (identification or verification) of an input sample. In this paper we present a method for computing the weights in a weighted sum fusion for score combinations, by means of a likelihood model. The maximum likelihood estimation is set as a linear programming problem. The scores are derived from a GMM classifier working on a different feature extractor. Our experimental results assesed the robustness of the system in front a changes on time (different sessions) and robustness in front a change of microphone. The improvements obtained were significantly better (error bars of two standard deviations) than a uniform weighted sum or a uniform weighted product or the best single classifier. The proposed method scales computationaly with the number of scores to be fussioned as the simplex method for linear programming.

Application of the Mutual Information Minimization to speaker recognition / identification improvement

In this paper we propose the inversion of nonlinear distortions in order to improve the recognition rates of a speaker recognizer system. We study the effect of saturations on the test signals, trying to take into account real situations where the training material has been recorded in a controlled situation but the testing signals present some mismatch with the input signal level (saturations). The experimental results for speaker recognition shows that a combination of several strategies can improve the recognition rates with saturated test sentences from 80% to 89.39%, while the results with clean speech (without saturation) is 87.76% for one microphone, and for speaker identification can reduce the minimum detection cost function with saturated test sentences from 6.42% to 4.15%, while the results with clean speech (without saturation) is 5.74% for one microphone and 7.02% for the other one.

On the Selection of Non-Invasive Methods Based on Speech Analysis Oriented to Automatic Alzheimer Disease Diagnosis

The work presented here is part of a larger study to identify novel technologies and biomarkers for early Alzheimer disease (AD) detection and it focuses on evaluating the suitability of a new approach for early AD diagnosis by non-invasive methods. The purpose is to examine in a pilot study the potential of applying intelligent algorithms to speech features obtained from suspected patients in order to contribute to the improvement of diagnosis of AD and its degree of severity. In this sense, Artificial Neural Networks (ANN) have been used for the automatic classification of the two classes (AD and control subjects). Two human issues have been analyzed for feature selection: Spontaneous Speech and Emotional Response. Not only linear features but also non-linear ones, such as Fractal Dimension, have been explored. The approach is non invasive, low cost and without any side effects. Obtained experimental results were very satisfactory and promising for early diagnosis and classification of AD patients.

On Automatic Diagnosis of Alzheimer's Disease based on Spontaneous Speech Analysis and Emotional Temperature

Alzheimer's disease is the most prevalent form of progressive degenerative dementia; it has a high socio-economic impact in Western countries. Therefore it is one of the most active research areas today. Alzheimer's is sometimes diagnosed by excluding other dementias, and definitive confirmation is only obtained through a post-mortem study of the brain tissue of the patient. The work presented here is part of a larger study that aims to identify novel technologies and biomarkers for early Alzheimer's disease detection, and it focuses on evaluating the suitability of a new approach for early diagnosis of Alzheimer’s disease by non-invasive methods. The purpose is to examine, in a pilot study, the potential of applying Machine Learning algorithms to speech features obtained from suspected Alzheimer sufferers in order help diagnose this disease and determine its degree of severity. Two human capabilities relevant in communication have been analyzed for feature selection: Spontaneous Speech and Emotional Response. The experimental results obtained were very satisfactory and promising for the early diagnosis and classification of Alzheimer’s disease patients.

A Robust Multiple Feature Approach To Endpoint Detection In Car Environment Based On Advanced Classifiers

In this paper we propose an endpoint detection system based on the use of several features extracted from each speech frame, followed by a robust classifier (i.e Adaboost and Bagging of decision trees, and a multilayer perceptron) and a finite state automata (FSA). We present results for four different classifiers. The FSA module consisted of a 4-state decision logic that filtered false alarms and false positives. We compare the use of four different classifiers in this task. The look ahead of the method that we propose was of 7 frames, which are the number of frames that maximized the accuracy of the system. The system was tested with real signals recorded inside a car, with signal to noise ratio that ranged from 6 dB to 30dB. Finally we present experimental results demonstrating that the system yields robust endpoint detection.

Design Methodology for Corrugated Metal Pipe Tiedowns, HR-332, Phase 1, 1993

Questionnaires were sent to transportation agencies in all 50 states in the U.S., to Puerto Rico, and all provinces in Canada asking about their experiences with uplift problems of - corrugated metal pipe (CMP). Responses were received from 52 agencies who reported 9 failures within the last 5 years. Some agencies also provided design standards for tiedowns to resist uplift. There was a wide variety in restraining forces used; for example for a pipe 6 feet in diameter, the resisting force ranged from 10 kips to 66 kips. These responses verified the earlier conclusion based on responses from Iowa county engineers that a potential uplift danger exists.when end restraint is not provided for CMP and that existing designs have an unclear theoretical or experimental basis. In an effort to develop more rational design standards, the longitudinal stiffness of three CMP ranging from 4 to 8 feet in diameter were measured in the laboratory. Because only three tests were conducted, a theoretical model to evaluate the stiffness of pipes of a variety of gages and corrugation geometries was also developed. The experimental results indicated a "stiffness" EI in the range of 9.11 x 10^5 k-in^2 to 34.43 x 10^5 k-in^2 for the three pipes with the larger diameter pipes having greater stiffness. The theoretical model developed conservatively estimates these stiffnesses.

Precast Concrete Panel Thickness for Epoxy-Coated Prestressing Strands, Volume 2, HR-353, 1994

This final report for Phase 1 of the research on epoxy-coated, prestressing strands in precast prestressed concrete (PC) panels has been published in two volumes. Volume 1--Technical Report contains the problem description, literature review, and survey results; descriptions of the test specimens, experimental tests, and analytical models; discussions of the analytical and experimental results; summary, conclusions, and recommendations; list of references; and acknowledgments. Volume 2--Supplemental Report contains additional information in the form of appendix material for Volume 1 on the questionnaires, strand forces, geometry of the specimens, concrete crack patterns that formed in the strand transfer length and strand development length specimens, concrete strains in the strand transfer length specimens, and load-point deflections and strand-slip measurements for the strand development length specimens. Appendix A contains the questionnaires that were sent to the design agencies and precast concrete producers. A summary of the results to the questions on the surveys are given as the number of respondents who provided the same answers and as paraphrased comments from the respondents. Appendix B contains graphs of strand force versus time, strand force versus temperature, and strand force versus strand cutting sequence for the concrete castings. Appendix C contains figures that show the location of each specimen in the prestress bed, the geometrical configurations for the strand transfer length (T-type) specimens and strand development length (D-type) specimens, and the concrete cracks that developed in some of the T-type specimens when they were prestressed. Appendix D contains figures that show the concrete cracks that developed in the D-type specimens during the strand development length tests. For each of these tests, the sequence of the failure for the specimen is specified. Appendix E contains graphs of concrete strain versus distance from the end of the T-type specimens that were instrumented with internal embedment strain gages. Appendix F contains graphs of load versus load-point deflection and load versus strand-slip for the strand development length tests of the D-type specimens.

Investigation of Plastic Pipes for Highway Applications, HR-373A, Phase II, 1997

It is generally accepted that high density polyethylene pipe (HDPE) performs well under live loads with shallow cover, provided the backfill is well compacted. Although industry standards require carefully compacted backfill, poor inspection and/or faulty construction may result in soils that provide inadequate restraint at the springlines of the pipes thereby causing failure. The objectives of this study were: 1) to experimentally define a lower limit of compaction under which the pipes perform satisfactorily, 2) to quantify the increase in soil support as compaction effort increases, 3) to evaluate pipe response for loads applied near the ends of the buried pipes, 4) to determine minimum depths of cover for a variety of pipes and soil conditions by analytically expanding the experimental results through the use of the finite element program CANDE. The test procedures used here are conservative especially for low-density fills loaded to high contact stresses. The failures observed in these tests were the combined effect of soil bearing capacity at the soil surface and localized wall bending of the pipes. Under a pavement system, the pipes' performance would be expected to be considerably better. With those caveats, the following conclusions are drawn from this study. Glacial till compacted to 50% and 80% provides insufficient support; pipe failureoccurs at surface contact stresses lower than those induced by highway trucks. On the other hand, sand backfill compacted to more than 110 pcf (17.3 kN/m3) is satisfactory. The failure mode for all pipes with all backfills is localized wall bending. At moderate tire pressures, i.e. contact stresses, deflections are reduced significantly when backfill density is increased from about 50 pcf (7.9 kN/m^3) to 90 pcf (14.1 kN/m^3). Above that unit weight, little improvement in the soil-pipe system is observed. Although pipe stiffness may vary as much as 16%, analyses show that backfill density is more important than pipe stiffness in controlling both deflections at low pipe stresses and at the ultimate capacity of the soil-pipe system. The rate of increase in ultimate strength of the system increases nearly linearly with increasing backfill density. When loads equivalent to moderate tire pressures are applied near the ends of the pipes, pipe deflections are slighly higher than when loaded at the center. Except for low density glacial till, the deflections near the ends are not excessive and the pipes perform satisfactorily. For contact stresses near the upper limit of truck tire pressures and when loaded near the end, pipes fail with localized wall bending. For flowable fill backfill, the ultimate capacity of the pipes is nearly doubled and at the upper limit of highway truck tire pressures, deflections are negligible. All pipe specimens tested at ambient laboratory room temperatures satisfied AASHTO minimum pipe stiffness requirements at 5% deflection. However, nearly all specimens tested at elevated pipe surface temperatures, approximately 122°F (50°C), failed to meet these requirements. Some HDPE pipe installations may not meet AASHTO minimum pipe stiffness requirements when installed in the summer months (i.e. if pipe surface temperatures are allowed to attain temperatures similar to those tested here). Heating of any portion of the pipe circumference reduced the load carrying capacity of specimens. The minimum soil cover depths, determined from the CANOE analysis, are controlled by the 5% deflection criterion. The minimum soil cover height is 12 in. (305 mm). Pipes with the poor silt and clay backfills with less than 85% compaction require a minimum soil cover height of 24 in. (610 mm). For the sand at 80% compaction, the A36 HDPE pipe with the lowest moment of inertia requires a minimum of 24 in. (610 mm) soil cover. The C48 HDPE pipe with the largest moment of inertia and all other pipes require a 12 in. (305 mm) minimum soil cover.

Risk aversion and embedding bias

In Selten (1967) ?Strategy Method,? the second mover in the game submits a complete strategy. This basic idea has been exported to nonstrategic experiments, where a participant reports a complete list of contingent decisions, one for each situation or state in a given sequence, out of which one and only one state, randomly selected, will be implemented.In general, the method raises the following concern. If S0 and S1 are two differentsequences of states, and state s is in both S0 and S1, would the participant make the same decision in state s when confronted with S0 as when confronted with S1? If not, the experimental results are suspect of suffering from an ?embedding bias.?We check for embedding biases in elicitation methods of Charles Holt and Susan Laury(Laury and Holt, 2000, and Holt and Laury, 2002), and of the present authors (Bosch-Dom?nech and Silvestre, 1999, 2002, 2006a, b) by appropriately chosen replications of the original experiments. We find no evidence of embedding bias in our work. But in Holt and Laury?s method participants tend to switch earlier to the riskier option when later pairs of lotteries are eliminated from the sequence, suggesting the presence of some embedding bias.

A new drug delivery system inhibits uveitis in an animal model after cataract surgery.

Cataract surgery is a common ocular surgical procedure consisting in the implantation of an artificial intraocular lens (IOL) to replace the ageing, dystrophic or damaged natural one. The management of postoperative ocular inflammation is a major challenge especially in the context of pre-existing uveitis. The association of the implanted IOL with a drug delivery system (DDS) allows the prolonged intraocular release of anti-inflammatory agents after surgery. Thus IOL-DDS represents an "all in one" strategy that simultaneously addresses both cataract and inflammation issues. Polymeric DDS loaded with two model anti-inflammatory drugs (triamcinolone acetonide (TA) and cyclosporine A (CsA)) were manufactured in a novel way and tested regarding their efficiency for the management of intraocular inflammation during the 3 months following surgery. The study involved an experimentally induced uveitis in rabbits. Experimental results showed that medicated DDS efficiently reduced ocular inflammation (decrease of protein concentration in aqueous humour, inflammatory cells in aqueous humour and clinical score). Additionally, more than 60% of the loading dose remained in the DDS at the end of the experiment, suggesting that the system could potentially cover longer inflammatory episodes. Thus, IOL-DDS were demonstrated to inhibit intraocular inflammation for at least 3 months after cataract surgery, representing a potential novel approach to cataract surgery in eyes with pre-existing uveitis.

Zinc adsorption in highly weathered soils

The objective of this work was to assess the effects of pH and ionic strength upon zinc adsorption, in three highly weathered variable charge soils. Adsorption isotherms were elaborated from batch adsorption experiments, with increasing Zn concentrations (0-80 mg L-1), and adsorption envelopes were constructed through soil samples reactions with 0.01, 0.1 and 1 mol L-1 Ca(NO3)2 solutions containing 5 mg L-1 of Zn, with an increasing pH value from 3 to 8. Driving force of reaction was quantified by Gibbs free energy and separation factor. Isotherms were C-, H- and L-type and experimental results were fitted to nonlinear Langmuir model. Maximum adsorption ranged from 59-810 mg kg-1, and Zn affinity was greater in subsoil (0.13-0.81 L kg-1) than in the topsoil samples (0.01-0.34 L kg-1). Zinc adsorption was favorable and spontaneous, and showed sharply increase (20-90%) in the 4-6 pH range. No effect of ionic strength was observed at pH values below 5, because specific adsorption mechanisms predominated in the 3-5 pH range. Above pH 5, and in subsoil samples, Zn was adsorbed by electrostatic mechanisms, since ionic strength effect was observed. Despite depth and ionic strength effects, Zn adsorption depends mainly on the pH.

Unraveling the "pressure effect" in nucleation

The influence of the pressure of a chemically inert carrier gas on the nucleation rate is one of the biggest puzzles in the research of gas-liquid nucleation. Experiments can show a positive effect, a negative effect, or no effect at all. The same experiment may show both trends for the same substance depending on temperature, or for different substances at the same temperature. We show how this ambiguous effect naturally arises from the competition of two contributions: nonisothermal effects and pressure-volume work. Our model clarifies seemingly contradictory experimental results and quantifies the variation of the nucleation ability of a substance in the presence of an ambient gas. Our findings are corroborated by molecular dynamics simulations and might have important implications since nucleation in experiments, technical applications, and nature practically always occurs in the presence of an ambient gas.

A Thermokinetic approach to radiative heat transfer at the nanoscale

Radiative heat exchange at the nanoscale presents a challenge for several areas due to its scope and nature. Here, we provide a thermokinetic description of microscale radiative energy transfer including phonon-photon coupling manifested through a non-Debye relaxation behavior. We show that a lognormal-like distribution of modes of relaxation accounts for this non-Debye relaxation behavior leading to the thermal conductance. We also discuss the validity of the fluctuation-dissipation theorem. The general expression for the thermal conductance we obtain fits existing experimental results with remarkable accuracy. Accordingly, our approach offers an overall explanation of radiative energy transfer through micrometric gaps regardless of geometrical configurations and distances.

Study of OSEM with different subsets in grating-based X-ray differential phase-contrast imaging.

Impressive developments in X-ray imaging are associated with X-ray phase contrast computed tomography based on grating interferometry, a technique that provides increased contrast compared with conventional absorption-based imaging. A new "single-step" method capable of separating phase information from other contributions has been recently proposed. This approach not only simplifies data-acquisition procedures, but, compared with the existing phase step approach, significantly reduces the dose delivered to a sample. However, the image reconstruction procedure is more demanding than for traditional methods and new algorithms have to be developed to take advantage of the "single-step" method. In the work discussed in this paper, a fast iterative image reconstruction method named OSEM (ordered subsets expectation maximization) was applied to experimental data to evaluate its performance and range of applicability. The OSEM algorithm with different subsets was also characterized by comparison of reconstruction image quality and convergence speed. Computer simulations and experimental results confirm the reliability of this new algorithm for phase-contrast computed tomography applications. Compared with the traditional filtered back projection algorithm, in particular in the presence of a noisy acquisition, it furnishes better images at a higher spatial resolution and with lower noise. We emphasize that the method is highly compatible with future X-ray phase contrast imaging clinical applications.

Loss of Prestress, Camber, and Deflection of Non-Composite and Composite Structures Using Different Weight Concretes, HR-137, 1970

General equations are presented for predicting loss of prestress and camber of both composite and non- composite prestressed concrete structures. Continuous time functins of all parameters needed to solve the equations are given, and sample results included. Computed prestress loss and camber are compared with experimental data for normal weight and lightweight concrete. Methods are also presented for predicting the effect of non-prestressed tension steel in reducing time-dependent loss of prestress and camber, and for the determination of short-time deflections of uncracked and cracked prestressed members. Comparisons with experimental results are indicated for these partially prestressed methods.