The goal of this thesis is to design and test techniques for unsupervised adaptation of speech recognizers on some audio data without any textual transcripts. A training set is prepared at first, and a baseline speech recognition system is trained. This sistem is used to transcribe some unseen data. We will experiment with an adaptation data selection process based on some speech transcript quality measurement. The system is re-trained on this new set than, and the accuracy is evaluated. Then we experiment with the amount of adaptation data.

archives@tulane.edu / Structural changes in the cytoskeleton during metastatic transformation make cancer cells more deformable, and recent experimental studies confirm a direct correlation between cell invasiveness and cell deformability. Several microfluidic approaches have recently developed to exploit this cellular property for high-throughput assessment of metastatic risk from small samples of patient’s blood. While demonstrating feasibility in the lab, these technologies often lack a solid theoretical foundation or do not show adequate sensitivity to cellular mechanical properties (“mechanotype”). The long-term goal of this project is to optimize microfluidic tests for metastatic risk assessment, including circulating tumor cell (CTC) isolation and mechanotype analysis, through predictive computational modeling. Specific aims of the presented study are 1) to expand the capability of our custom computational algorithm for viscoelastic cell deformation and migration to simulate cell sorting and CTC isolation in channels with complex geometry, including channels with pillars and bifurcations, and 2) to demonstrate the capability of our algorithm to optimize microfluidic methods for cancer cell mechanotype measurement. / 1 / Scott J. Hymel

Common wiki applications lack possibilities to structure the relationships between wiki pages. This paper presents a semantic wiki prototype named SHAWN that allows modelling concepts and their relationships within a wiki environment. One goal of this prototype is to keep concept creation very simple. Yet, entering relationship data instantaneously gratifies the user with enhanced navigational means on the wiki. The engine supports simple semantic queries upon the emergent model. A challenge is to accommodate a self-explaining query interface for these ontologies.

No description available.

Models are widely used for communication and documentation purposes. They also tend to be used as parameters for code generation. Because these models have to be complete, consistent and correct we have to support modellers in keeping their models clean. The modeller should have the choice to select the preferred view and the needed level of detail for his modelling and model maintenance tasks. This paper proposes a model architecture named AMABULO for a model driven development process of business logic for information systems. The model architecture consists of a meta model, corresponding visual diagrams and an interchange format. With the use of AMABULO the development process for business logic is supported from analysis until code generation.

Background: As the business world faces significant changes to adapt to the rapidly increasing sustainability demand, they also face significant changes to their business core. The European Commission established a circular action plan in 2015, pressuring businesses operating in the EU area to work to implement sustainability in their operations proactively.    Purpose: The aim is to understand how companies select, develop and incorporate sustainable factors in their business model. The gap in the existing literature was identified to be within the field of illustrating the process of sustainable business model innovation. The authors aim to discuss the existing literature and provide new insights on the chosen subject.   Method: A multiple case study was designed with a qualitative research approach to align the study with the purpose. The case study built on data collection through semi-structured interviews, letting the interviewees carry the most significant part of the conversation. This was beneficial since the authors adopted an interpretive research paradigm, highlighting the importance of individuals' perceptions of the process.    Conclusion: The results show that the decisions that drive the implementation of sustainable business model innovation are often profit-driven but can also be driven by morale and ethics, governmental regulations, and stakeholder pressure. Both internal and external resistance can be expected when transforming a company towards sustainability. Therefore, broad acceptance amongst shareholders and stakeholders will enable a transformation, which can partly be achieved through a robust organizational identity that ties sustainability to it.

El texto completo de este trabajo no está disponible en el Repositorio Académico UPC por restricciones de la casa editorial donde ha sido publicado. / This research paper proposes a low-cost telepresence technological model focused on primary education. Its aim is to give students a new resource/communication channel for classes, which would be used when they cannot attend school due to health problems that do not affect their learning process. This solution seeks students to not be passive listeners during a session, but that they interact with their classmates and teachers during class. To validate the model, a telepresence platform based on WebRTC was developed. It was tested in three schools in different geographical areas belonging to socioeconomic sector C, collecting data from the students who tested the tool, as well as from classmates, teachers, and parents. / Revisión por pares

A microcomputer simulation was developed as a tool for analyzing the dairy replacement enterprise. The simulation was constructed using a spreadsheet, and equations were developed using stepwise regression procedures. The simulation predicted BW, DMI, and fixed and variable costs for each week of a heifer's life from birth to calving. After calving, milk yield, feed costs, and fixed costs were predicted for first lactation. Variation was estimated for each predicted variable, thus enabling normal distribution of predicted values. The simulation was used to analyze profitability of various growth rate scenarios and marginal costs associated with changing feed costs, heat detection efficiency, death loss, and abortion rate. For the growth analysis, six scenarios were evaluated: 1) normal growth from 5 wk to calving, 2) Accelerated growth from 5 wk to calving, 3) slow growth from 5 wk to calving, 4) normal Growth from 5 wk to 14 mo and accelerated growth from 14 mo to calving, 5) accelerated growth from 5 wk to 14 mo and control growth from 14 mo to calving, and 6) slow growth from 5 wk to 14 mo and accelerated growth from 14 mo to calving. Average daily gain from birth to calving was 0.78, 0.90, 0.62, 0.78,0.75, and 0.80 kg/d, and age at calving was 25.1, 23.1, 27.4, 23.1, 23.0, and 23.1 mo,respectively. Total rearing cost from birth to calving was 1246, 1220, 1275, 1148, 1148, and 1138 $/heifer, and net profit through first lactation was 399, 407, 319, 441, 432, and 463 $/heifer, respectively. Results suggest modest growth rates from birth to calving (0.75 to 0.80 kg/d) with reduced first calving age <24 mo) is most desirable, and delayed calving (>24 mo) is costly and merits higher growth rates with earlier breeding. Increasing feed costs, death loss at birth through weaning, or abortion rate one percentage point increased rearing costs 7.33, 2.40, and 9.10 $/heifer. Improving heat detection efficiency one percentage point reduced rearing costs $2.80/heifer. For the heat detection analysis, the relationship between age at first calving and total rearing costs was -584.38 + 73.49 x calving age in mo (R-squared = 0.97), for ages at first calving from 24.4 to 26.6 mo. Results of this research agree with field observations that managers should strive for early calving (<24 mo) and modest growth rates (0.75 to 0.80 kg/d) to maximize profitability of the replacement enterprise. In addition, death loss, abortion rate, and heat detection efficiency are variables that a manager must control to minimize heifer rearing costs. / Ph. D.

In this study, we numerically explore methods of coupling sensitivity analysis to the reduced model in order to increase the accuracy of a proper orthogonal decomposition (POD) basis across a wider range of parameters. Various techniques based on polynomial interpolation and basis alteration are compared. These techniques are performed on a 1-dimensional reaction-diffusion equation and 2-dimensional incompressible Navier-Stokes equations solved using the finite element method (FEM) as the full scale model. The expanded model formed by expanding the POD basis with the orthonormalized basis sensitivity vectors achieves the best mixture of accuracy and computational efficiency among the methods compared. / Master of Science

The purpose of this paper is to present a maintenance study focused on total productive maintenance (TPM) and reliability-centered maintenance (RCM). Its approach is based on the first pillars of TPM, preventive and autonomous maintenance, as well as the FMEA analysis of RCM for maintenance analysis, which was conducted in this study. The implementation of TPM was successful in that various preventive maintenance (PM) policies assigned to the assets were implemented and it was demonstrated that TPM application in the construction industry could reduce the excessive accumulation of maintenance with the same effective optimization, and with support from RCM analysis and its heavy equipment systems analysis. Excessive corrective maintenance accounts for high investment and delay rates in work times of the assigned project. Traditional methods of availability guarantee, such as reactive or routine maintenance, are insufficient to satisfy a heavy equipment maintenance plan; therefore, what is called for is the systematic application of RCM and TMP because they allow the selection and application of effective PM tasks. An approach that develops and thoroughly analyzes the strategies of continuous corrective and PM is used with an atmosphere of uncertainty and with operational data limited by criticism. Results show a 90% improvement in availability. / Revisión por pares