In this thesis, we address the action retrieval and the object category segmentation problems by directly optimizing application specific performance measures using deep learning. Most deep learning methods are designed to optimize simple loss functions (e.g., cross-entropy or hamming loss). These loss functions are suitable for applications where the performance of the application is measured by overall accuracy. But for many applications, the overall accuracy is not an appropriate performance measure. For example, applications like action retrieval often use the area under the Receiver Operating Characteristic curve (ROC curve) to measure the performance of a retrieval algorithm. Likewise, in object category segmentation from images, the intersection-over-union (IoU) is the standard performance measure. In this thesis, we propose approaches to directly optimize these complex performance measures in deep learning framework. / October 2016

Sėkmingas muzikinio rašto įsisavinimas tai ne tik sėkmingo mokymosi groti fortepijonu veiksnys, pateiktas įdomiai ir žaismingai dar vaikystėje, tai gali būti ir būdas sudominti bei pritraukti vaikus mokytis muzikos. Ketverių metų vaikų amžius pasižymi aukštu imlumu bei įsiminimu. Praleistas šis periodas yra pagrindinė priežastis, dėl kurios nėra realizuojami įgimti gabumai ir polinkiai. Tai galima traktuoti, kaip aktualią pedagoginę problemą. Siekiant ją išspręsti reikėtų išnaudoti šį, patį imliausią vaikų amžių. Remiantis žaidimo principu, tokio amžiaus vaikams galima suteikti nemažai muzikinio rašto pažinimo žinių. Tyrimo objektas – muzikinio raštingumo įtaka pradiniam mokymo groti fortepijonu etapui. Tyrimo tikslas – atskleisti muzikinio raštingumo svarbą optimizuojant paruošiamąjį grojimo fortepijonu etapą. Pirmajame darbo skyriuje analizuojama žmogaus muzikinio ugdymo samprata, bei jo ypatumai, antrajame skyriuje aptariami keturių – penkerių metų vaikų amžiaus tarpsnių ypatumai, taip pat analizuojami sėkmingo muzikinio rašto įsisavinimo ir fortepijono pažinimo veiksniai, o trečiajame – aprašomas ugdomasis eksperimentas ir pateikiama „Muzikos rašto ir fortepijono pažinimo“ programa. Ši programa – optimalus būdas žaismingai supažindinti vaikus su muzikinio rašto pagrindais ir įtraukti juos į muzikos pasaulį. Žinios vaikų nepervargina, yra įdomios ir nenuobodžios. Jas pateikiant vaikai nesupranta, kad vyksta mokymo procesas. Programa patikrinta vaikų darželyje „Žiogelis... [toliau žr. visą tekstą] / Mastering of musical literacy is not just a factor of successful learning to play the grand piano, when presented in an engaging and playful manner already in childhood it may be a method for engaging and attracting the children to learn to play the grand piano. Children at age of four demonstrate high susceptibility and memorizing. If the latter period is missed it may become the main reason for non-realization of abilities and aptitudes. This might be treated as a relevant pedagogical problem. In pursuance of solving the latter one should make use of the most susceptible age of children. On the basis of the principle of game, children of such age may be awarded a wealth of knowledge on musical literacy cognition. The object of the research – impact of musical literacy upon primary stage of teaching to play the grand piano. The aim of the research – to reveal significance of musical literacy for optimizing preparatory stage of playing the grand piano. The first part of the thesis analyses man’s conception of musical education as well as peculiarities thereof, in the second part the peculiarities of a period of age of four-five as well as the factors of successful mastering of musical writing and cognition of the grand piano are discussed, and finally, in the third part educational research is described and program on “Cognition of Musical Writing and the Grand Piano” is presented. The above program is an optimal way to introduce the children the basics of musical... [to full text]

No description available.

Intermediate storage system optimization is an essential aspect which this report tends to look upon and putting effort on improving operating efficiency then reducing the capital cost of batch/ semi-continuous or flow production process by looking at the extent of estimating the figures of boxes that are made to be in the intermediate storage system through evaluations. Additionally, intermediate storage can reduce the effects of process parameter variations, such as recipe inaccuracies and operator errors, in which process operations are particularly visible. In a channel of operations, where there are many amount of shift that are inevitable to pass through in order to complete production process to get a finished product. Hence, longer operating horizon, process variations must include a measurable parameter, either shorter or longer to avoid failure of meeting product specification and limitation of meeting end users specification. Intermediate storage system as a back-up function in case of some sudden failure within the production line In this thesis, it is notable to make an amendment, as a way of knowing the accuracy of boxes of bearing work-piece that will be maintaining a certain intermediate storage system to feed the next shift of workstation in a continuous process with accurate or known estimated number of work-piece in a role, until the final product is been made. Hence, in this research, there are many theory that were adopted to give a clear view of what will be obtainable in the whole research both in empirical finding to have a definite result. To achieve the above goal this thesis followed a steps to fulfil its objective. Steps 1 and 2 were to know the strategic plans as well as operating methods through theory also, to know the production line balancing to have a view of time measured through the theory. Step 3 and 4 adoptions of theory and empirical finding to calculate, productivity through the source of cycle time, as well as facility plan and to give analysis of intermediate storage system, in optimization of production line to make a work flow achievable at the end, to meet the customers demand.

Adaptive Just-in-Time compilers employ multiple techniques to concentrate compilation efforts in the most promising spots of the application, balancing tight compilation budgets with an appropriate level of code quality. Some compiler researchers propose that Just-in-Time compilers should benefit from method-specific compilation strategies. These strategies can be discovered through machine-learning techniques, where a compilation strategy is tailored to a method based on the method's characteristics. This thesis investigates the use of Support Vector Machines in Testarossa, a commercial Just-in-Time compiler employed in the IBM J9 Java Virtual Machine. This new infrastructure allows Testarossa to explore numerous compilation strategies, generating the data needed for training such models. The infrastructure also integrates Testarossa to learned models that predict which compilation strategy balances code quality and compilation effort, on a per-method basis. The thesis also presents the results of an extensive experimental evaluation of the infrastructure and compares these results with the performance of the original Testarossa.

This thesis deals with the optimization of stocks of medicines in the pharmacy. The goal is to find a connection when dealing with inventory optimization drugs. First is the optimization of inventories as such. It is also discussed the situation in the pharmaceutical market in different ATC groups in the Czech Republic. Selected methods is processed inventory optimization of specific drugs in the existing pharmacy equipment.

Subject of this dissertation will be the introduction of information systém into planning and optimizing of transport with regard to cost saving, improving the quality of distribution services for customers within its process, improving the efficiency of work of trafic dispatcher and bringing the professionalism into their business, logistic process planning and fleet optimizing.

En una coyuntura de constante cambio mundial, junto con un ambiente económico y político complejo para el Perú, el panorama para el sector de seguros vehiculares local se ve impactado desde 2017 con un decrecimiento del 5% según la SBS. Momentos complicados y cambiantes como los actuales, obligan a las compañias de seguros a replantear sus estrategias de corto y largo plazo. El presente trabajo de investigación tomó como base la metodología Lean Service, información de casos de éxito y tecnologías de la industria de seguros global, con el objetivo de analizar y optimizar los procesos de la unidad de seguros vehiculares de una compañía de seguros. Las estrategias de mejora de procesos no sólo se enfocan a nivel operativo, sino estratégico con el fin de que la compañía obtenga un valor diferenciador sostenible en el mercado, puede adoptar un enfoque customer centricity y sea más eficiente, para así lograr la mayor rentabilidad posible. La propuesta de implementación ha sido trabajada bajo la metodología Roadmaping, con el fin de obtener una secuencia lógica para el desarrollo de las iniciativas planteadas. Finalmente, cada mejora ha sido valorizada y se realizaron proyecciones de impacto en los principales indicadores de la unidad de seguros vehiculares de la compañía. / At a time of constant global change, along with a complex economic and political environment in Peru, the outlook for the local vehicle insurance sector has been impacted since 2017 with a 5% decrease according to the SBS. Complicated and unstable periods like the current ones, force insurance companies to rethink their short- and long-term strategies. This research paper was based on the Lean Service methodology, information on successful implementations and technologies of the global insurance industry, with the aim of analyzing and optimizing the processes of the vehicle insurance unit of an insurance company. Process improvement strategies are not only focused at an operational level, but also at a strategic one, so that the company obtains a sustainable differentiating value in the market, adopt a customer centricity approach and be more efficient, in order to achieve the highest possible profitability. The implementation proposal has been worked based on the Roadmapping methodology, in order to obtain a logical sequence for the development of the proposed initiatives. Finally, each improvement has been valued and impact projections were made on the main indicators of the company's vehicle insurance unit. / Trabajo de investigación

Efficiency of current cell cultures producing biological products is limited due to accumulation of the product and waste on the cells. A previous work aims at assisting bioengineers in solving this problem by introducing a new set of cells which possess the ability to connect to one another and thus form a network. Once created, the network of cells can help the initial cellules by removing their waste and product as well as nourishing them. Our project explores a technique to be able to create the best network of such cells; hence maximizing the amount of metabolic product in the cell culture.

A nation-wide survey of hardwood sawmills was conducted in the fall of 1999. The objectives of the survey were to determine the differences between adopters and non-adopters of scanning and optimizing technology, identify the company expectations of scanning and optimizing technology, and model the adoption decision process for future scanning and optimizing technology. These objectives were chosen because timely information was not available on the hardwood sawmill industry, and even less was known about the overall state of technology with the industry. The survey consisted of a mail questionnaire which was sent to over 2000 hardwood sawmills. The questionnaire was used to collect demographic, equipment, and preference scale information on the hardwood sawmill industry. The second part of this project used the Analytic Hierarchy Process to model the adoption decision process for future scanning and optimizing technology in hardwood sawmills. Data was collected through personal interviews with two hardwood sawmill groups including adopters and non-adopters of advanced scanning and optimizing technology. The interviewee rated the importance of the decision factors in the adoption decision process. They also rated the influence of four sawmill departments on the adoption decision process. The results from the mail survey found that the average yearly lumber production was 7.6 million board feet per sawmill. The most common type of scanning and optimizing technology, headrig optimization, was only in use by 27 percent of the responding mills. Advanced scanning and optimizing technology such as edger-optimizers and trimmer-optimizers were only in use by 10 percent and 5 percent of the respondents respectively. Adoption decision factors for scanning and optimizing technology were rated. Improved raw material recovery and increased lumber revenues were the two most highly rated factors. Accuracy of grading was the most highly rated factor for automated grading systems. The adoption decision model found that production related issues were most important in the decision process and that the production department was the most influential of the sawmill departments. Overall, scanning and optimizing technology adoption within the hardwood sawmill industry is low. For those that have adopted advanced scanning and optimizing technology, production issues were the driving factors. / Ph. D.