No description available.

Den 1 september 2020 trädde den konsoliderade versionen av Boverkets Byggregler BBR 29 i kraft. En stor del av skillnaden från föregående version var det ökade kravet på energihushållning. Det nya kravet som fastställts innebär att U-medelvärdet är sänkt från 0,4 W/m2K till 0,3 W/m2K. Detta är ett mått på hur väl en byggnad är isolerad. Företaget Kontio expanderade till Sverige från Finland för snart fyra år sedan. De säljer framförallt fritidsboende och småhus byggda av egentillverkade timmerstockar med korslimmad arktisk furu i olika dimensioner. Deras hus säljs på den finska marknaden, men i Sverige har endast vissa timmerdimensioner klarat det gamla kravet från BBR 28. Med det nya kravet måste klimatskärmen anpassas till den nya kravbilden. Tak och golv är färdigprojekterade med låga U-värden, därför finns den största möjligheten för anpassning i väggkonstruktionerna. Med hjälp av givna förutsättningar togs ett antal väggkonstruktioner fram för beräkning av U-värdet. Dessa gav upphov till olika typlösningar som varierar med, förutom väggkonstruktionen, fönstertyper och minskade fönsterareor. Framräknat U-värde för väggkonstruktion låg till grund för beräkning av U-medelvärde tillsammans med U-värde för fönster, golv och tak. Ändringar i väggkonstruktioner kan resultera i en förekomst av fuktproblem. Därför testas dessa med hjälp av en Glaser-tabell som undersöker genom stationära beräkningar huruvida det föreligger någon risk i framtagna väggkonstruktioner. Detta projekt analyserar även för- och nackdelar med timmer i furu och dess värmelagrande funktion som beror på en hög värmekapacitet. Med utgångspunkt i litteraturstudien analyseras och utreds var i konstruktionen man med störst fördel placerar den massiva timmerstrukturen i förhållande till värmeisolering. Alla väggkonstruktioner behövde göras om utom 275 där endast fönstertyp ändrades till de med lägre U-värde. Vid anpassning av väggkonstruktioner med timmerdimension 135 och 205 krävdes mer omfattande åtgärder. Genom att isolera väggarna, ändra fönstertyper och minska fönsterareor kunde kravet på U-medelvärde nås på olika sätt. Ingen av dessa vägguppbyggnader uppvisade tecken på fukttekniska problem. / On September 1, 2020, the consolidated version of the National Board of Housing, Building and Planning's Building Rules BBR 29 took effect. A large part of the difference from the previous version was the increased demand for energy management. The new requirement that has been established means that the mean U-value has been reduced from 0.4 W/m2K to 0.3 W/m2K. This is a measurement of how well a building is insulated. Almost four years ago, the company Kontio expanded to Sweden from Finland. They mainly sell holiday homes and living houses built from self-made logs with cross-laminated arctic pine in various dimensions. Their houses are sold on the Finnish market, but in Sweden only certain timber dimensions have met the old requirement from BBR 28. With the new BBR 29, the building shell must be adapted to the new requirement. Ceilings and floors are pre-designed with low U-values and for this reason further improvements are best made in the wall construction. Using the given conditions, a number of wall constructions were developed for calculating the U-value. These gave rise to different type solutions that vary with, in addition to the wall construction, window types and reduced window areas. Calculated U-value for wall construction was the basis for calculating the mean U-value together with the U-value for windows, floors and ceilings. Changes in wall constructions can result in the occurrence of moisture problems, therefore these walls are tested using a Glaser-table which examines, by stationary calculations, whether there is any risk in the adapted constructions. This project also analyzes the advantages and disadvantages of pine timber and its heat storage function due to its high heat capacity. Based on the literature study, it is analyzed and investigated where in the construction the massive timber structure is placed with the greatest advantage in relation to the thermal insulation. All wall constructions needed to be adapted except 275 where only the window type was changed to those with a lower U-value. When adapting wall constructions with timber dimensions 135 and 205, more extensive measures were required. By insulating the walls, changing window types and reducing window areas, the requirement for the mean U-value could be achieved in various ways. None of these wall constructions showed signs of moisture technical problems.

An increase in demand for sawed wood products is the reason why Norra Timber has to expand their facility at Sävar Sawmill. During the expansion, a new log sorting will be implemented which can sort timber with different characteristics and qualities compared to the current log sorting. The placement of the new log sorting and the logistics after the construction has not been decided. Therefore, the purpose of this thesis is to suggest a design of the new timber yard which includes placement of the new log sorting, log piles and roads to minimize travel distance for tractors between the log sorting and the saw. The thesis project started with examining the opportunities and challenges that exist in timber logistics through a literature study. Afterwards, the current logistics at Sävar Sawmill were studied. Relevant data was collected to create different layout suggestions and to determine which optimization model that should be used. The placement of the log sorting was determined after the data collection, together with a steering group at Sävar Sawmill. Three layout suggestions were created after the placement of the new log sorting and were restricted to the current asphalted surface for timber handling. All layout suggestions should have a storage capacity with the volumes 15 000 m3f, 20 000 m3f and 35 000 m3f. The unit of volume m3f, describes the solid volume of timber without bark. The assignment problem with the Hungarian algorithm was used to optimize the placement of log classes on the timber yard. Thereafter transport distances and driving hours were calculated with three different production levels. The results show that all timber fits in the timber yard for the three layout suggestions when the storage levels are 15 000 m3f and 20 000 m3f, where an optimal solution is found. In the beginning it was not possible to find an optimal solution when the storage levels increased to 35 000 m3f. To find a solution, one constraint was relaxed that allowed the log piles to become overfilled. Norra Timber did not believe a solution at 35 000 m3f existed and therefore accepted the relaxation. Layout suggestion 1 had the least amount of overfilled log piles and layout suggestion 2 had the most. There was a significant difference for the travel distance per year between the layout suggestions when the storage level was low compared to high. Layout suggestion 1 had the longest travel distance per year and layout suggestion 3 had the shortest for all the production levels. The positions and the number of log piles will be the same regardless the production levels, but the travel distance will increase when the production level increases. The number of driving hours is calculated from the travel distances to point out the differences in working hours for the tractor drivers. Layout suggestion 3 will be proposed to use in the future based on the result from the optimization model. The layout suggestion 3 has significant shorter transport distance for the two lower storage levels compared to the other suggestions. The tractor drivers also appreciated that layout suggestion 3 had more available log piles and that the placement of the log piles will ease their daily work and save many working hours. / En ökad efterfrågan på sågade trävaror gör att Norra Timber är i behov av att expandera sin nuvarande anläggning på Sävar såg. Vid expansionen kommer en ny timmersortering installeras som kan sortera timret på flera egenskaper och kvaliteter jämfört med den befintliga. Varken placeringen av den nya timmersorteringen eller hur logistiken ska fungera efter expansionen är bestämt. Syftet med examensarbetet är att ta fram ett förslag på hur Norra Timber ska designa området med placering av den nya timmersorteringen, vältplatser och vägar för att minimera transportavstånd för traktorerna mellan timmersorteringen och sågen.  Examensarbetet inleddes med att undersöka vilka möjligheter och utmaningar som finns inom timmerlogistik genom en litteraturstudie. Logistiken på Sävar såg studerades sedan och varje aktivitet förklarades i en nulägesbeskrivning. Relevant fakta samlades in för att kunna skapa olika designförslag och avgöra vilken optimeringsmodell som skulle användas. Efter datainsamling bestämdes placeringen av den nya timmersorteringen tillsammans med en projektgrupp på Sävar såg. Utifrån den nya timmersorteringens placering skapades tre olika designförslag som var begränsade till den asfalterade ytan vid den nuvarande timmerhanteringen. Alla designförslag skulle klara av en lagernivå på 15 000 m3f, 20 000 m3f och 35 000 m3f. Volymenheten m3f står för den fasta volymen timmer utan bark. För att optimera var timmerklasserna skulle placeras i vältor på timmerplanen användes tillordningsproblemet med Hungarian algoritm där indatat anpassades efter lager- och produktionsnivå. Transportavstånd och körtimmar räknas sedan ut för alla designförslag med tre olika produktionsnivåer.   Resultatet visar att när lagernivåerna är 15 000 m3f och 20 000 m3f får allt timmer plats på timmerplanen i alla designförslag och modellen hittar en optimal lösning. När lagernivån ökade till 35 000 m3f kunde inte en optimal lösning hittas till en början. För att hitta en lösning relaxerades ett bivillkor vilket gjorde att överfulla vältor tilläts. Norra Timber trodde inte en lösning vid 35 000 m3f existerade och accepterade därför relaxationen. Minst antal överfulla vältor hade designförslag 1 och designförslag 2 hade högst antal. Designförslagens transportavstånd per år skiljer sig mycket mellan den lägsta lagernivån jämfört med den högsta. Designförslag 1 har längst avstånd och designförslag 3 har kortast avstånd för alla lagernivåer. Oavsett vilken produktionsnivå sågen har kommer vältplaceringarna att vara samma för ett designförslag där endast den totala körsträckan ökar. Antal körtimmar beräknas utifrån transportavstånd och presenteras för att visa på skillnaderna mellan traktorförarnas arbetstid. Utifrån optimeringsmodellens resultat rekommenderas Norra Timber att designa den nya timmerplanen på Sävar såg efter designförslag 3. Förslaget har signifikant kortare transportavstånd vid de två lägre lagernivåerna jämfört med de andra två designförslagen. Traktorförarna uppskattade också att förslaget hade fler vältplatser och vältornas placering som kommer underlätta deras vardagliga arbete och spara in många arbetstimmar.

In last years Hidden Markov Models (HMMs) have received significant attention in the task off-line handwritten text recognition (HTR). As in automatic speech recognition (ASR), HMMs are used to model the probability of an observation sequence, given its corresponding text transcription. However, in contrast to what happens in ASR, in HTR there is no standard set of local features being used by most of the proposed systems. In this thesis we propose the use of raw binary pixels as features, in conjunction with models that deal more directly with the binary data. In particular, we propose the use of Bernoulli HMMs (BHMMs), that is, conventional HMMs in which Gaussian (mixture) distributions have been replaced by Bernoulli (mixture) probability functions. The objective is twofold: on the one hand, this allows us to better modeling the binary nature of text images (foreground/background) using BHMMs. On the other hand, this guarantees that no discriminative information is filtered out during feature extraction (most HTR available datasets can be easily binarized without a relevant loss of information). In this thesis, all the HMM theory required to develop a HMM based HTR toolkit is reviewed and adapted to the case of BHMMs. Specifically, we begin by defining a simple classifier based on BHMMs with Bernoulli probability functions at the states, and we end with an embedded Bernoulli mixture HMM recognizer for continuous HTR. Regarding the binary features, we propose a simple binary feature extraction process without significant loss of information. All input images are scaled and binarized, in order to easily reinterpret them as sequences of binary feature vectors. Two extensions are proposed to this basic feature extraction method: the use of a sliding window in order to better capture the context, and a repositioning method in order to better deal with vertical distortions. Competitive results were obtained when BHMMs and proposed methods were applied to well-known HTR databases. In particular, we ranked first at the Arabic Handwriting Recognition Competition organized during the 12th International Conference on Frontiers in Handwriting Recognition (ICFHR 2010), and at the Arabic Recognition Competition: Multi-font Multi-size Digitally Represented Text organized during the 11th International Conference on Document Analysis and Recognition (ICDAR 2011). In the last part of this thesis we propose a method for training BHMM classifiers using In last years Hidden Markov Models (HMMs) have received significant attention in the task off-line handwritten text recognition (HTR). As in automatic speech recognition (ASR), HMMs are used to model the probability of an observation sequence, given its corresponding text transcription. However, in contrast to what happens in ASR, in HTR there is no standard set of local features being used by most of the proposed systems. In this thesis we propose the use of raw binary pixels as features, in conjunction with models that deal more directly with the binary data. In particular, we propose the use of Bernoulli HMMs (BHMMs), that is, conventional HMMs in which Gaussian (mixture) distributions have been replaced by Bernoulli (mixture) probability functions. The objective is twofold: on the one hand, this allows us to better modeling the binary nature of text images (foreground/background) using BHMMs. On the other hand, this guarantees that no discriminative information is filtered out during feature extraction (most HTR available datasets can be easily binarized without a relevant loss of information). In this thesis, all the HMM theory required to develop a HMM based HTR toolkit is reviewed and adapted to the case of BHMMs. Specifically, we begin by defining a simple classifier based on BHMMs with Bernoulli probability functions at the states, and we end with an embedded Bernoulli mixture HMM recognizer for continuous HTR. Regarding the binary features, we propose a simple binary feature extraction process without significant loss of information. All input images are scaled and binarized, in order to easily reinterpret them as sequences of binary feature vectors. Two extensions are proposed to this basic feature extraction method: the use of a sliding window in order to better capture the context, and a repositioning method in order to better deal with vertical distortions. Competitive results were obtained when BHMMs and proposed methods were applied to well-known HTR databases. In particular, we ranked first at the Arabic Handwriting Recognition Competition organized during the 12th International Conference on Frontiers in Handwriting Recognition (ICFHR 2010), and at the Arabic Recognition Competition: Multi-font Multi-size Digitally Represented Text organized during the 11th International Conference on Document Analysis and Recognition (ICDAR 2011). In the last part of this thesis we propose a method for training BHMM classifiers using In last years Hidden Markov Models (HMMs) have received significant attention in the task off-line handwritten text recognition (HTR). As in automatic speech recognition (ASR), HMMs are used to model the probability of an observation sequence, given its corresponding text transcription. However, in contrast to what happens in ASR, in HTR there is no standard set of local features being used by most of the proposed systems. In this thesis we propose the use of raw binary pixels as features, in conjunction with models that deal more directly with the binary data. In particular, we propose the use of Bernoulli HMMs (BHMMs), that is, conventional HMMs in which Gaussian (mixture) distributions have been replaced by Bernoulli (mixture) probability functions. The objective is twofold: on the one hand, this allows us to better modeling the binary nature of text images (foreground/background) using BHMMs. On the other hand, this guarantees that no discriminative information is filtered out during feature extraction (most HTR available datasets can be easily binarized without a relevant loss of information). In this thesis, all the HMM theory required to develop a HMM based HTR toolkit is reviewed and adapted to the case of BHMMs. Specifically, we begin by defining a simple classifier based on BHMMs with Bernoulli probability functions at the states, and we end with an embedded Bernoulli mixture HMM recognizer for continuous HTR. Regarding the binary features, we propose a simple binary feature extraction process without significant loss of information. All input images are scaled and binarized, in order to easily reinterpret them as sequences of binary feature vectors. Two extensions are proposed to this basic feature extraction method: the use of a sliding window in order to better capture the context, and a repositioning method in order to better deal with vertical distortions. Competitive results were obtained when BHMMs and proposed methods were applied to well-known HTR databases. In particular, we ranked first at the Arabic Handwriting Recognition Competition organized during the 12th International Conference on Frontiers in Handwriting Recognition (ICFHR 2010), and at the Arabic Recognition Competition: Multi-font Multi-size Digitally Represented Text organized during the 11th International Conference on Document Analysis and Recognition (ICDAR 2011). In the last part of this thesis we propose a method for training BHMM classifiers using discriminative training criteria, instead of the conventionalMaximum Likelihood Estimation (MLE). Specifically, we propose a log-linear classifier for binary data based on the BHMM classifier. Parameter estimation of this model can be carried out using discriminative training criteria for log-linear models. In particular, we show the formulae for several MMI based criteria. Finally, we prove the equivalence between both classifiers, hence, discriminative training of a BHMM classifier can be carried out by obtaining its equivalent log-linear classifier. Reported results show that discriminative BHMMs clearly outperform conventional generative BHMMs. / Giménez Pastor, A. (2014). Bernoulli HMMs for Handwritten Text Recognition [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/37978

京都大学 / 新制・課程博士 / 博士(工学) / 甲第25259号 / 工博第5218号 / 新制||工||1995(附属図書館) / 京都大学大学院工学研究科都市社会工学専攻 / (主査)教授 林 為人, 教授 村田 澄彦, 教授 福山 英一 / 学位規則第4条第1項該当 / Doctor of Agricultural Science / Kyoto University / DFAM

The aim of this project was to investigate a potential connection between a molecule's capacity to penetrate the blood-brain barrier and its electrostatic potential on the surface of the molecule. A molecule's capacity to penetrate the blood-brain barrier is quantified by their log BB value. This was done by creating a set of 58 molecules, divided into a test set and a calibration set, in Gaussian 16. The molecules were then optimized and their quantum chemical properties were calculated by using HS-95. By using stepwise regression in both Matlab and Excel these different parameters were then used to search for a correlation between the parameters and experimental log BB values. This resulted in an equation with 10 variables with an adjusted R2 value of 0,6456. When tested against the test set the mean fault was 0,1443. Outliers were then identified and removed and stepwise regression was executed once more. This resulted in an equation with 10 variables with an adjusted R2 0,8749. When this was tested against the test set the mean fault was 0,0798. The results showed that the variables that were important were the electrostatic potential, surface area and volume.

This dissertation is composed of three articles describing novel approaches for analysis and modeling using multivariate skew-normal and skew-t distributions in econometrics and environmetrics. In the first article we introduce the Heckman selection-t model. Sample selection arises often as a result of the partial observability of the outcome of interest in a study. In the presence of sample selection, the observed data do not represent a random sample from the population, even after controlling for explanatory variables. Heckman introduced a sample-selection model to analyze such data and proposed a full maximum likelihood estimation method under the assumption of normality. The method was criticized in the literature because of its sensitivity to the normality assumption. In practice, data, such as income or expenditure data, often violate the normality assumption because of heavier tails. We first establish a new link between sample-selection models and recently studied families of extended skew-elliptical distributions. This then allows us to introduce a selection-t model, which models the error distribution using a Student’s t distribution. We study its properties and investigate the finite-sample performance of the maximum likelihood estimators for this model. We compare the performance of the selection-t model to the Heckman selection model and apply it to analyze ambulatory expenditures. In the second article we introduce a family of multivariate log-skew-elliptical distributions, extending the list of multivariate distributions with positive support. We investigate their probabilistic properties such as stochastic representations, marginal and conditional distributions, and existence of moments, as well as inferential properties. We demonstrate, for example, that as for the log-t distribution, the positive moments of the log-skew-t distribution do not exist. Our emphasis is on two special cases, the log-skew-normal and log-skew-t distributions, which we use to analyze U.S. precipitation data. Many commonly used statistical methods assume that data are normally distributed. This assumption is often violated in practice which prompted the development of more flexible distributions. In the third article we describe two such multivariate distributions, the skew-normal and the skew-t, and present commands for fitting univariate and multivariate skew-normal and skew-t regressions in the statistical software package Stata.

本文主要是在探討三個變數所構成之三維列聯表中，兩兩有序類別變數間的關係，而衡量的標準，我們則採用「兩兩變數所構成之二維列聯表中，相鄰兩列與相鄰兩行所求計出的相對成敗比（local odds ratios）」。在三維列聯表的資料架構下，我們可分別就固定某一變數水準之下兩個有序變數彼此間的「條件關係」，以及三個有序類別變數彼此兩兩間的「部分關係」，建構其各自的三維關係模型，並進行參數估計。此外，我們也提供必要的電腦程式，並舉出實例，加以說明。 / In analyzing a three-way contingency table with three ordinal variables, we can use association models suggested in Goodman (1979) to study the association between each pair of ordinal variables. The association was measured in terms of the local odds ratios formed from adjacent rows and adjacent columns of the cross-classification. This article investigates in great details the conditional association models and the partial association models for three-way cross-classifications. In addition, issues on estimating the para-meters in these two kinds of association models are discussed, and computer programs are provided. Some of the applications are illustrated.

In an automotive infotainment system, while analyzing bug reports, developers have to spend significant time on reading log messages and trying to locate anomalous behavior before identifying its root cause. The log messages need to be viewed in a Traceviewer tool to read in a human readable form and have to be extracted to text files by applying manual filters in order to further analyze the behavior. There is a need to evaluate machine learning/data mining methods which could potentially assist in error analysis. One such method could be learning patterns for “normal” messages. “Normal” could even mean that they contain keywords like “exception”, “error”, “failed” but are harmless or not relevant to the bug that is currently analyzed. These patterns could then be applied as a filter, leaving behind only truly anomalous messages that are interesting for analysis. A successful application of the filter would reduce the noise, leaving only a few “anomalous” messages. After evaluation of the researched candidate algorithms, two algorithms namely GSP and FP Growth were found useful and thus implemented together in a prototype. The prototype implementation overall includes processes like pre-processing, creation of input, executing algorithms, creation of training set and analysis of new trace logs. Execution of prototype resulted in reducing manual effort thus achieving the objective of this thesis work.

目前許多網路應用程式（web application），遵循三層式開發架構，並藉由「會談」（session）營造出具有狀態機制的環境。雖然會談能將流程串起，其重要性無庸置疑，卻因分層的關係，造成想要在任一記錄點上，取得會談資訊的困難。 本研究的目的，即在嘗試以「剖面導向程式設計」（Aspect-Oriented Programming）為基礎，希望能打造一個簡單易用的工具，將「記錄」（log）這種非功能性，屬於橫切面考量的需求，暫時排除在開發階段外，到事後再以挑選的方式，更輕鬆地加入或修改。更重要的是，將會談資訊中的使用者帳號，或可資識別的序號，內嵌在每行記錄裏，因此稱為「具會談概念的記錄」。讓記錄內容的設計者，不需再為了如何封裝、傳遞、拆解而浪費時間。因為記錄與系統開發是獨立的，即使上線後，臨時性的記錄修改需求，也不至於對既有的穩定與效能，造成重大影響，或不可復原的災難性結果。 / Logging is an important part to application development for purposes such as debugging and auditing. While there are many good frameworks and utilities for assisting logging in application development, it is still not an easy task for the logging action to provide enough information for those purposes. Specifically, the user account information in particular or the session information in general is not easy to obtain when we perform logging for a designated backend operation such as database query. The reasons are twofold. Firstly, applications do not use user accounts, but have their own accounts to access database. Secondly, the three-tiered structure of an application makes it difficult to thread the user accounts to the backend for logging. This thesis presents an aspect-based utility for Java-based Web applications that enables developer to generate customizable logs with session information, including user accounts. Our tool employs two aspects: one predefined aspect for collecting session information and the other is generic one for producing user-defined logs at designated application points. Specifically, the tool takes a wizard-based approach that guides the user to customize the second aspect for its own target points in a user-friendly manner.