Modeling Efficient Classification as a Process of Confidence Assessment and Delegation

In visual object detection and recognition, classifiers have two interesting characteristics: accuracy and speed. Accuracy depends on the complexity of the image features and classifier decision surfaces. Speed depends on the hardware and the computational effort required to use the features and decision surfaces. When attempts to increase accuracy lead to increases in complexity and effort, it is necessary to ask how much are we willing to pay for increased accuracy. For example, if increased computational effort implies quickly diminishing returns in accuracy, then those designing inexpensive surveillance applications cannot aim for maximum accuracy at any cost. It becomes necessary to find trade-offs between accuracy and effort. We study efficient classification of images depicting real-world objects and scenes. Classification is efficient when a classifier can be controlled so that the desired trade-off between accuracy and effort (speed) is achieved and unnecessary computations are avoided on a per input basis. A framework is proposed for understanding and modeling efficient classification of images. Classification is modeled as a tree-like process. In designing the framework, it is important to recognize what is essential and to avoid structures that are narrow in applicability. Earlier frameworks are lacking in this regard. The overall contribution is two-fold. First, the framework is presented, subjected to experiments, and shown to be satisfactory. Second, certain unconventional approaches are experimented with. This allows the separation of the essential from the conventional. To determine if the framework is satisfactory, three categories of questions are identified: trade-off optimization, classifier tree organization, and rules for delegation and confidence modeling. Questions and problems related to each category are addressed and empirical results are presented. For example, related to trade-off optimization, we address the problem of computational bottlenecks that limit the range of trade-offs. We also ask if accuracy versus effort trade-offs can be controlled after training. For another example, regarding classifier tree organization, we first consider the task of organizing a tree in a problem-specific manner. We then ask if problem-specific organization is necessary.

Redox-linked proton transfer by cytochrome c oxidase

The respiratory chain is found in the inner mitochondrial membrane of higher organisms and in the plasma membrane of many bacteria. It consists of several membrane-spanning enzymes, which conserve the energy that is liberated from the degradation of food molecules as an electrochemical proton gradient across the membrane. The proton gradient can later be utilized by the cell for different energy requiring processes, e.g. ATP production, cellular motion or active transport of ions. The difference in proton concentration between the two sides of the membrane is a result of the translocation of protons by the enzymes of the respiratory chain, from the negatively charged (N-side) to the positively charged side (P-side) of the lipid bilayer, against the proton concentration gradient. The endergonic proton transfer is driven by the flow of electrons through the enzymes of the respiratory chain, from low redox-potential electron donors to acceptors of higher potential, and ultimately to oxygen. Cytochrome c oxidase is the last enzyme in the respiratory chain and catalyzes the reduction of dioxygen to water. The redox reaction is coupled to proton transport across the membrane by a yet unresolved mechanism. Cytochrome c oxidase has two proton-conducting pathways through which protons are taken up to the interior part of the enzyme from the N-side of the membrane. The K-pathway transfers merely substrate protons, which are consumed in the process of water formation at the catalytic site. The D-pathway transfers both substrate protons and protons that are pumped to the P-side of the membrane. This thesis focuses on the role of two conserved amino acids in proton translocation by cytochrome c oxidase, glutamate 278 and tryptophan 164. Glu278 is located at the end of the D-pathway and is thought to constitute the branching point for substrate and pumped protons. In this work, it was shown that although Glu278 has an important role in the proton transfer mechanism, its presence is not an obligatory requirement. Alternative structural solutions in the area around Glu278, much like the ones present in some distantly related heme-copper oxidases, could in the absence of Glu278 support the formation of a long hydrogen-bonded water chain through which proton transfer from the D-pathway to the catalytic site is possible. The other studied amino acid, Trp164, is hydrogen bonded to the ∆-propionate of heme a3 of the catalytic site. Mutation of this amino acid showed that it may be involved in regulation of proton access to a proton acceptor, a pump site, from which the proton later is expelled to the P-side of the membrane. The ion pair that is formed by the ∆-propionate of heme a3 and arginine 473 is likely to form a gate-like structure, which regulates proton mobility to the P-side of the membrane. The same gate may also be part of an exit path through which water molecules produced at the catalytically active site are removed towards the external side of the membrane. Time-resolved optical and electrometrical experiments with the Trp164 to phenylalanine mutant revealed a so far undetected step in the proton pumping mechanism. During the A to PR transition of the catalytic cycle, a proton is transferred from Glu278 to the pump site, located somewhere in the vicinity of the ∆-propionate of heme a3. A mechanism for proton pumping by cytochrome c oxidase is proposed on the basis of the presented results and the mechanism is discussed in relation to some relevant experimental data. A common proton pumping mechanism for all members of the heme-copper oxidase family is moreover considered.

Neurotrophins and neuronal plasticity in the action of antidepressants and morphine

Neuronal plasticity is a well characterized phenomenon in the developing and adult brain. It refers to capasity of a single neuron to modify morphology, synaptic connections and activity. Neuronal connections and capacity for plastic events are compromised in several pathological disorders, such as major depression. In addition, neuronal atrophy has been reported in depressive patients. Neurotrophins are a group of secretory proteins functionally classified as neuronal survival factors. Neurotrophins, especially brain derived neurotrophic factor (BDNF), have also been associated with promoting neuronal plasticity in dysfunctional neuronal networks. Chronic antidepressant treatment increases plastic events including neurogenesis and arborization and branching of neurites in distinct brain areas, such as the hippocampus. One suggested mode of action is where the antidepressants elevate the synaptic levels of BDNF thus further activating several signaling cascades via trkB-receptor. In our studies we have tried to clarify the mechanisms of action for antidepressants and to resolve the role of BDNF in this process. We found that chronic antidepressant treatment increases amount of markers of neuronal plasticity in both hippocampus and in the medial prefrontal cortex, both of which are closely linked to the etiology of major depression. Secondary actions of antidepressants include rapid activation of the trkB receptor followed by a phosphorylation of transcription factor CREB. In addition, activation of CREB by phosphorylation appears responsible for the regulation of the expression of the BDNF gene. Using transgenic mice we found that BDNF-induced trkB-mediated signaling proved crucial for the behavioral effects of antidepressants in the forced swimming test and for the survival of newly-born neurons in the adult hippocampus. Antidepressants not only increased neurogenesis in the adult hippocampus but also elevated the turnover of hippocampal neurons. During these studies we also discovered that another trkB ligand, NT-4, is involved in morphine-mediated anti-nociception and tolerance. These results present a novel role for trkB-mediated signaling in plastic events present in the opioid system. This thesis evaluates neuronal plasticity and trkB as a target for future antidepressant treatments.

Microbial diversity in the municipal composting process and development of detection methods

Composting refers to aerobic degradation of organic material and is one of the main waste treatment methods used in Finland for treating separated organic waste. The composting process allows converting organic waste to a humus-like end product which can be used to increase the organic matter in agricultural soils, in gardening, or in landscaping. Microbes play a key role as degraders during the composting-process, and the microbiology of composting has been studied for decades, but there are still open questions regarding the microbiota in industrial composting processes. It is known that with the traditional, culturing-based methods only a small fraction, below 1%, of the species in a sample is normally detected. In recent years an immense diversity of bacteria, fungi and archaea has been found to occupy many different environments. Therefore the methods of characterising microbes constantly need to be developed further. In this thesis the presence of fungi and bacteria in full-scale and pilot-scale composting processes was characterised with cloning and sequencing. Several clone libraries were constructed and altogether nearly 6000 clones were sequenced. The microbial communities detected in this study were found to differ from the compost microbes observed in previous research with cultivation based methods or with molecular methods from processes of smaller scale, although there were similarities as well. The bacterial diversity was high. Based on the non-parametric coverage estimations, the number of bacterial operational taxonomic units (OTU) in certain stages of composting was over 500. Sequences similar to Lactobacillus and Acetobacteria were frequently detected in the early stages of drum composting. In tunnel stages of composting the bacterial community comprised of Bacillus, Thermoactinomyces, Actinobacteria and Lactobacillus. The fungal diversity was found to be high and phylotypes similar to yeasts were abundantly found in the full-scale drum and tunnel processes. In addition to phylotypes similar to Candida, Pichia and Geotrichum moulds from genus Thermomyces and Penicillium were observed in tunnel stages of composting. Zygomycetes were detected in the pilot-scale composting processes and in the compost piles. In some of the samples there were a few abundant phylotypes present in the clone libraries that masked the rare ones. The rare phylotypes were of interest and a method for collecting them from clone libraries for sequencing was developed. With negative selection of the abundant phylotyps the rare ones were picked from the clone libraries. Thus 41% of the clones in the studied clone libraries were sequenced. Since microbes play a central role in composting and in many other biotechnological processes, rapid methods for characterization of microbial diversity would be of value, both scientifically and commercially. Current methods, however, lack sensitivity and specificity and are therefore under development. Microarrays have been used in microbial ecology for a decade to study the presence or absence of certain microbes of interest in a multiplex manner. The sequence database collected in this thesis was used as basis for probe design and microarray development. The enzyme assisted detection method, ligation-detection-reaction (LDR) based microarray, was adapted for species-level detection of microbes characteristic of each stage of the composting process. With the use of a specially designed control probe it was established that a species specific probe can detect target DNA representing as little as 0.04% of total DNA in a sample. The developed microarray can be used to monitor composting processes or the hygienisation of the compost end product. A large compost microbe sequence dataset was collected and analysed in this thesis. The results provide valuable information on microbial community composition during industrial scale composting processes. The microarray method was developed based on the sequence database collected in this study. The method can be utilised in following the fate of interesting microbes during composting process in an extremely sensitive and specific manner. The platform for the microarray is universal and the method can easily be adapted for studying microbes from environments other than compost.

Cytomegalovirus in the Process of Chronic Allograft Nephropathy

Cytomegalovirus (CMV) is a major cause of morbidity, costs and even mortality in organ transplant recipients. CMV may also enhance the development of chronic allograft nephropathy (CAN), which is the most important cause of graft loss after kidney transplantation. The evidence for the role of CMV in chronic allograft nephropathy is somewhat limited, and controversial results have also been reported. The aim of this study was to investigate the role of CMV in the pathogenesis of CAN. Material for the purpose of this study was available from altogether 70 kidney transplant recipients who received a kidney transplant between the years 1992-2000. CMV infection was diagnosed with pp65 antigenemia test or by viral culture from blood, urine, or both. CMV proteins were demonstrated in the kidney allograft biopsies by immunohistochemisrty and CMV-DNA by in situ hybridization. Cytokines, adhesion molecules, and growth factors were demonstrated from allograft biopsies by immunohistochemistry, and from urinary samples by ELISA-methods. CMV proteins were detectable in the 6-month protocol biopsies from 18/41 recipients with evidence of CMV infection. In the histopathological analysis of the 6-month protocol biopsies, presence of CMV in the allograft together with a previous history of acute rejection episodes was associated with increased arteriosclerotic changes in small arterioles. In urinary samples collected during CMV infection, excretion of TGF-β was significantly increased. In recipients with increased urinary excretion of TGF-β, increased interstitial fibrosis was recorded in the 6- month protocol biopsies. In biopsies taken after an active CMV infection, CMV persisted in the kidney allograft in 17/48 recipients, as CMV DNA or antigens were detected in the biopsies more than 2 months after the last positive finding in blood or urine. This persistence was associated with increased expression of TGF-β, PDGF, and ICAM-1 and with increased vascular changes in the allografts. Graft survival and graft function one and two years after transplantation were reduced in recipients with persistent intragraft CMV. Persistent intragraft CMV infection was also a risk factor for reduced graft survival in Cox regression analysis, and an independent risk factor for poor graft function one and two years after transplantation in logistic regression analysis. In conclusion, these results show that persistent intragraft CMV infection is detrimental to kidney allografts, causing increased expression of growth factors and increased vascular changes, leading to reduced graft function and survival. Effective prevention, diagnosis and treatment of CMV infections may a major factor in improving the long term survival of kidney allograft.

First Measurement of the Ratio of Branching Fractions B(Lambda_b to Lambda_c mu nu)/B(Lambda_b to Lambda_c pi)

The analysis uses data from an integrated luminosity of approximately 172 pb-1 of ppbar collisions at sqrt(s)=1.96 TeV, collected with the CDF II detector at the Fermilab Tevatron. The Lambda_b and B0 relative branching fractions are measured to be: B(Lambda_b to Lambda_c+ mu nu)/B(Lambda_b to Lambda_c+ pi) = 16.6 +- 3.0 (stat) +- 1.0 (syst) +2.6 -3.4 (PDG) +- 0.3 (EBR), B(B0 to D+ mu nu)/B(B0 to D+ pi) = 9.9 +- 1.0 (stat) +- 0.6 (syst) +- 0.4 (PDG) +- 0.5 (EBR), B(B0 to D*+ mu nu)/B(B0 to D*+ pi) = 16.5 +- 2.3 (stat) +- 0.6 (syst) +- 0.5 (PDG) +- 0.8 (EBR) This article also presents measurements of the branching fractions of four new Lambda_b semileptonic decays: Lambda_b to Lambda_c(2595)+ mu nu, Lambda_b to Lambda_c(2625)+ mu nu, Lambda_b to Sigma_c(2455)0 pi mu nu, Lambda_b to Sigma_c(2455)++ pi mu nu, relative to the branching fraction of the Lambda_b to Lambda_c mu nu decay. Finally, the transverse-momentum distribution of Lambda_b baryons produced in p-pbar collisions is measured and found to be significantly different from that of B0 mesons.

First observation of the decay Bs --> Ds K and measurement of the ratio of branching fractions Br(Bs --> DsK)/Br(Bs --> Ds pi)

A combined mass and particle identification fit is used to make the first observation of the decay Bs --> Ds K and measure the branching fraction of Bs --> Ds K relative to Bs --> Ds pi. This analysis uses 1.2 fb^-1 integrated luminosity of pbar-p collisions at sqrt(s) = 1.96 TeV collected with the CDF II detector at the Fermilab Tevatron collider. We observe a Bs --> Ds K signal with a statistical significance of 8.1 sigma and measure Br(Bs --> Ds K)/Br(Bs --> Ds pi) = 0.097 +- 0.018(stat) +- 0.009(sys).

Measurements of branching fraction ratios and CP asymmetries in B±→DCPK± decays in hadron collisions

We reconstruct B+/- -> D K+/- decays in a data sample collected by the CDF II detector at the Tevatron collider corresponding to 1 fb-1 of integrated luminosity. We select decay modes where the D meson decays to either K- pi+ (flavor eigenstate) or K- K+, pi- pi+ (CP-even eigenstates), and measure the direct CP asymmetry A_CP+ = 0.39 +/- 0.17(stat) +/- 0.04(syst), and the double ratio of CP-even to flavor eigenstate branching fractions R_CP+ = 1.30 +/- 0.24(stat) +/- 0.12(syst). These measurements will improve the determination of the CKM angle gamma. They are performed here for the first time using data from hadron collisions.

First measurement of the ratio of branching fractions B(Λb0→Λc+μ-ν̅ μ)/B(Λb0→Λc+π-)

This article presents the first measurement of the ratio of branching fractions B(Λb0→Λc+μ-ν̅ μ)/B(Λb0→Λc+π-). Measurements in two control samples using the same technique B(B̅ 0→D+μ-ν̅ μ)/B(B̅ 0→D+π-) and B(B̅ 0→D*(2010)+μ-ν̅ μ)/B(B̅ 0→D*(2010)+π-) are also reported. The analysis uses data from an integrated luminosity of approximately 172  pb-1 of pp̅ collisions at √s=1.96  TeV, collected with the CDF II detector at the Fermilab Tevatron. The relative branching fractions are measured to be B(Λb0→Λc+μ-ν̅ μ)/B(Λb0→Λc+π-)=16.6±3.0(stat)±1.0(syst)+2.6/-3.4(PDG)±0.3(EBR), B(B̅ 0→D+μ-ν̅ μ)/B(B̅ 0→D+π-)= 9.9±1.0(stat)±0.6(syst)±0.4(PDG)±0.5(EBR), and B(B̅ 0→D*(2010)+μ-ν̅ μ)/B(B̅ 0→D*(2010)+π-)=16.5±2.3(stat)± 0.6(syst)±0.5(PDG)±0.8(EBR). The uncertainties are from statistics (stat), internal systematics (syst), world averages of measurements published by the Particle Data Group or subsidiary measurements in this analysis (PDG), and unmeasured branching fractions estimated from theory (EBR), respectively. This article also presents measurements of the branching fractions of four new Λb0 semileptonic decays: Λb0→Λc(2595)+μ-ν̅ μ, Λb0→Λc(2625)+μ-ν̅ μ, Λb0→Σc(2455)0π+μ-ν̅ μ, and Λb0→Σc(2455)++π-μ-ν̅ μ, relative to the branching fraction of the Λb0→Λc+μ-ν̅ μ decay. Finally, the transverse-momentum distribution of Λb0 baryons produced in pp̅ collisions is measured and found to be significantly different from that of B̅ 0 mesons, which results in a modification in the production cross-section ratio σΛb0/σB̅ 0 with respect to the CDF I measurement.

First Observation of B̅ s0→Ds±K∓ and Measurement of the Ratio of Branching Fractions B(B̅ s0→Ds±K∓)/B(B̅ s0→Ds+π-)

A combined mass and particle identification fit is used to make the first observation of the decay B̅ s0→Ds±K∓ and measure the branching fraction of B̅ s0→Ds±K∓ relative to B̅ s0→Ds+π-. This analysis uses 1.2  fb-1 integrated luminosity of pp̅ collisions at √s=1.96  TeV collected with the CDF II detector at the Fermilab Tevatron collider. We observe a B̅ s0→Ds±K∓ signal with a statistical significance of 8.1σ and measure B(B̅ s0→Ds±K∓)/B(B̅ s0→Ds+π-)=0.097±0.018(stat)±0.009(syst).

First Observation of B̅ s0→Ds±K∓ and Measurement of the Ratio of Branching Fractions B(B̅ s0→Ds±K∓)/B(B̅ s0→Ds+π-)

A combined mass and particle identification fit is used to make the first observation of the decay Bs --> Ds K and measure the branching fraction of Bs --> Ds K relative to Bs --> Ds pi. This analysis uses 1.2 fb^-1 integrated luminosity of pbar-p collisions at sqrt(s) = 1.96 TeV collected with the CDF II detector at the Fermilab Tevatron collider. We observe a Bs --> Ds K signal with a statistical significance of 8.1 sigma and measure Br(Bs --> Ds K)/Br(Bs --> Ds pi) = 0.097 +- 0.018(stat) +- 0.009(sys).

A model Integrating the Facilities Management Process with the Building End User’s Business Process (ProFacil)

The ProFacil model is a generic process model defined as a framework model showing the links between the facilities management process and the building end user’s business process. The purpose of using the model is to support more detailed process modelling. The model has been developed using the IDEF0 modelling method. The ProFacil model describes business activities from the generalized point of view as management-, support-, and core processes and their relations. The model defines basic activities in the provision of a facility. Examples of these activities are “operate facilities”, “provide new facilities”, “provide re-build facilities”, “provide maintained facilities” and “perform dispose of facilities”. These are all generic activities providing a basis for a further specialisation of company specific FM activities and their tasks. A facilitator can establish a specialized process model using the ProFacil model and interacting with company experts to describe their company’s specific processes. These modelling seminars or interviews will be done in an informal way, supported by the high-level process model as a common reference.

A Formalised Model of the Information and Materials Handling Activities in the Construction Process

A model of the information and material activities that comprise the overall construction process is presented, using the SADT activity modelling methodology. The basic model is further refined into a number of generic information handling activities such as creation of new information, information search and retrieval, information distribution and person-to-person communication. The viewpoint could be described as information logistics. This model is then combined with a more traditional building process model, consisting of phases such as design and construction. The resulting two-dimensional matrix can be used for positioning different types of generic IT-tools or construction specific applications. The model can thus provide a starting point for a discussion of the application of information and communication technology in construction and for measurements of the impacts of IT on the overall process and its related costs.