New kernel functions and learning methods for text and data mining

Recent advances in machine learning methods enable increasingly the automatic construction of various types of computer assisted methods that have been difficult or laborious to program by human experts. The tasks for which this kind of tools are needed arise in many areas, here especially in the fields of bioinformatics and natural language processing. The machine learning methods may not work satisfactorily if they are not appropriately tailored to the task in question. However, their learning performance can often be improved by taking advantage of deeper insight of the application domain or the learning problem at hand. This thesis considers developing kernel-based learning algorithms incorporating this kind of prior knowledge of the task in question in an advantageous way. Moreover, computationally efficient algorithms for training the learning machines for specific tasks are presented. In the context of kernel-based learning methods, the incorporation of prior knowledge is often done by designing appropriate kernel functions. Another well-known way is to develop cost functions that fit to the task under consideration. For disambiguation tasks in natural language, we develop kernel functions that take account of the positional information and the mutual similarities of words. It is shown that the use of this information significantly improves the disambiguation performance of the learning machine. Further, we design a new cost function that is better suitable for the task of information retrieval and for more general ranking problems than the cost functions designed for regression and classification. We also consider other applications of the kernel-based learning algorithms such as text categorization, and pattern recognition in differential display. We develop computationally efficient algorithms for training the considered learning machines with the proposed kernel functions. We also design a fast cross-validation algorithm for regularized least-squares type of learning algorithm. Further, an efficient version of the regularized least-squares algorithm that can be used together with the new cost function for preference learning and ranking tasks is proposed. In summary, we demonstrate that the incorporation of prior knowledge is possible and beneficial, and novel advanced kernels and cost functions can be used in algorithms efficiently.

Design Space Exploration for MPSoC Architectures

Multiprocessor system-on-chip (MPSoC) designs utilize the available technology and communication architectures to meet the requirements of the upcoming applications. In MPSoC, the communication platform is both the key enabler, as well as the key differentiator for realizing efficient MPSoCs. It provides product differentiation to meet a diverse, multi-dimensional set of design constraints, including performance, power, energy, reconfigurability, scalability, cost, reliability and time-to-market. The communication resources of a single interconnection platform cannot be fully utilized by all kind of applications, such as the availability of higher communication bandwidth for computation but not data intensive applications is often unfeasible in the practical implementation. This thesis aims to perform the architecture-level design space exploration towards efficient and scalable resource utilization for MPSoC communication architecture. In order to meet the performance requirements within the design constraints, careful selection of MPSoC communication platform, resource aware partitioning and mapping of the application play important role. To enhance the utilization of communication resources, variety of techniques such as resource sharing, multicast to avoid re-transmission of identical data, and adaptive routing can be used. For implementation, these techniques should be customized according to the platform architecture. To address the resource utilization of MPSoC communication platforms, variety of architectures with different design parameters and performance levels, namely Segmented bus (SegBus), Network-on-Chip (NoC) and Three-Dimensional NoC (3D-NoC), are selected. Average packet latency and power consumption are the evaluation parameters for the proposed techniques. In conventional computing architectures, fault on a component makes the connected fault-free components inoperative. Resource sharing approach can utilize the fault-free components to retain the system performance by reducing the impact of faults. Design space exploration also guides to narrow down the selection of MPSoC architecture, which can meet the performance requirements with design constraints.

Purification and fractionation by nanotiltration in dairy and sugar and sweetener industry applications

Nanotiltration is a membrane separation method known for its special characteristic of rejecting multivalent ions and passing monovalent ions. Thus, it is commonly applied with dilute aqueous solutions in partial salt removal, like in drinking water production. The possibilities of nanofiltration have been studied and the technique applied in a wide branch of industries, e.g. the pulp and paper, the textile and the chemical processing industry. However, most present applications and most of the potential applications studied involve dilute solutions, the permeating stream being generally water containing monovalent salts. In this study nanotiltration is investigated more as a fractionation method. A well-known application in the dairy industry is concentration and partial salt removal from whey. Concentration and partial demineralization is beneficial for futher processing of whey as whey concentrates are used e.g. in baby foods. In the experiments of this study nanotiltration effectively reduced the monovalent salts in the whey concentrate. The main concern in this application is lactose leakage into the permeate. With the nanofiltration membranes used the lactose retentions were practically ? 99%. Another dairy application studied was the purification and reuse of cleaning solutions. This is an environmentally driven application. An 80% COD reduction by nanofiltration was observed for alkaline cleaning-in-place solution. Nanofiltration is not as commonly applied in the sugar and sweeteners industry as in the dairy industry. In this study one potential application was investigated, namely xylose purification from hemicellulose hydrolyzate. Xylose is raw material for xylitol production. Xylose separation from glucose was initially studied with xylose-glucose model solutions. The ability of nanofiltration to partially separate xylose into the permeate from rather concentrated xylose-glucose solutions (10 w-% and 30 w-%) became evident. The difference in size between xylose and glucose molecules according to any size measure is small, e.g. the Stokes diameter of glucose is 0.73 nm compared to 0.65 nm for xylose. In further experiments, xylose was purified into nanoliltration permeate from a hemicellulose hydrolyzate solution. The xylose content in the total solids was increased by 1.4—1.7 fold depending on temperature, pressure and feed composition.

Investigation and data adaptive analysis of CARS spectra

In this diploma work advantages of coherent anti-Stokes Raman scattering spectrometry (CARS) and various methods of the quantitative analysis of substance structure with its help are considered. The basic methods and concepts of the adaptive analysis are adduced. On the basis of these methods the algorithm of automatic measurement of a scattering strip size of a target component in CARS spectrum is developed. The algorithm uses known full spectrum of target substance and compares it with a CARS spectrum. The form of a differential spectrum is used as a feedback to control the accuracy of matching. To exclude the influence of a background in CARS spectra the differential spectrum is analysed by means of its second derivative. The algorithm is checked up on the simulated simple spectra and on the spectra of organic compounds received experimentally.

Trigger and data link system for CMS Resistive Plate Chambers at the LHC accelerator

The purpose of the work was to realize a high-speed digital data transfer system for RPC muon chambers in the CMS experiment on CERN’s new LHC accelerator. This large scale system took many years and many stages of prototyping to develop, and required the participation of tens of people. The system interfaces to Frontend Boards (FEB) at the 200,000-channel detector and to the trigger and readout electronics in the control room of the experiment. The distance between these two is about 80 metres and the speed required for the optic links was pushing the limits of available technology when the project was started. Here, as in many other aspects of the design, it was assumed that the features of readily available commercial components would develop in the course of the design work, just as they did. By choosing a high speed it was possible to multiplex the data from some the chambers into the same fibres to reduce the number of links needed. Further reduction was achieved by employing zero suppression and data compression, and a total of only 660 optical links were needed. Another requirement, which conflicted somewhat with choosing the components a late as possible was that the design needed to be radiation tolerant to an ionizing dose of 100 Gy and to a have a moderate tolerance to Single Event Effects (SEEs). This required some radiation test campaigns, and eventually led to ASICs being chosen for some of the critical parts. The system was made to be as reconfigurable as possible. The reconfiguration needs to be done from a distance as the electronics is not accessible except for some short and rare service breaks once the accelerator starts running. Therefore reconfigurable logic is extensively used, and the firmware development for the FPGAs constituted a sizable part of the work. Some special techniques needed to be used there too, to achieve the required radiation tolerance. The system has been demonstrated to work in several laboratory and beam tests, and now we are waiting to see it in action when the LHC will start running in the autumn 2008.

Structural and operational optimisation : applications in energy systems

In this work mathematical programming models for structural and operational optimisation of energy systems are developed and applied to a selection of energy technology problems. The studied cases are taken from industrial processes and from large regional energy distribution systems. The models are based on Mixed Integer Linear Programming (MILP), Mixed Integer Non-Linear Programming (MINLP) and on a hybrid approach of a combination of Non-Linear Programming (NLP) and Genetic Algorithms (GA). The optimisation of the structure and operation of energy systems in urban regions is treated in the work. Firstly, distributed energy systems (DES) with different energy conversion units and annual variations of consumer heating and electricity demands are considered. Secondly, district cooling systems (DCS) with cooling demands for a large number of consumers are studied, with respect to a long term planning perspective regarding to given predictions of the consumer cooling demand development in a region. The work comprises also the development of applications for heat recovery systems (HRS), where paper machine dryer section HRS is taken as an illustrative example. The heat sources in these systems are moist air streams. Models are developed for different types of equipment price functions. The approach is based on partitioning of the overall temperature range of the system into a number of temperature intervals in order to take into account the strong nonlinearities due to condensation in the heat recovery exchangers. The influence of parameter variations on the solutions of heat recovery systems is analysed firstly by varying cost factors and secondly by varying process parameters. Point-optimal solutions by a fixed parameter approach are compared to robust solutions with given parameter variation ranges. In the work enhanced utilisation of excess heat in heat recovery systems with impingement drying, electricity generation with low grade excess heat and the use of absorption heat transformers to elevate a stream temperature above the excess heat temperature are also studied.

E-thesis repository – processes and data

Poster at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

Targeted Annotation of Scientific Literature and Data Resources in Invenio Digital Libraries

Presentation at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

Building Dynamic Data Centers for Fast Delivery of New Data and Data Updates

Poster at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

Apply of energy storages in electricity distribution networks and reserve capacity applications

The Thesis is dedicated to development of an operative tool to support decision making of battery energy storages implementation in distribution networks. The basics of various battery technologies, their perspectives and challenges are represented in the Thesis. Mathematical equations that describe economic effect from battery energy storage installation are offered. The main factors that influence profitability of battery settings have been explored and mathematically defined. Mathematical model and principal trends of battery storage profitability under an impact of the major factors are determined. The meaning of annual net value was introduced to show the difference between savings and required costs. The model gives a clear vision for dependencies between annual net value and main factors. Proposals for optimal network and battery characteristics are suggested.

Role of service and data reuse in enterprises

Digitalization has been predicted to change the future as a growing range of non-routine tasks will be automated, offering new kinds of business models for enterprises. Serviceoriented architecture (SOA) provides a basis for designing and implementing welldefined problems as reusable services, allowing computers to execute them. Serviceoriented design has potential to act as a mediator between IT and human resources, but enterprises struggle with their SOA adoption and lack a linkage between the benefits and costs of services. This thesis studies the phenomenon of service reuse in enterprises, proposing an ontology to link different kinds of services with their role conceptually as a part of the business model. The proposed ontology has been created on the basis of qualitative research conducted in three large enterprises. Service reuse has two roles in enterprises: it enables automated data sharing among human and IT resources, and it may provide cost savings in service development and operations. From a technical viewpoint, the ability to define a business problem as a service is one of the key enablers for achieving service reuse. The research proposes two service identification methods, first to identify prospective services in the existing documentation of the enterprise and secondly to model the services from a functional viewpoint, supporting service identification sessions with business stakeholders.

Blue and Uv-Emitting Upconversion Nanoparticles: Synthesis and (Bio) Analytical Applications.

Rare-earth based upconverting nanoparticles (UCNPs) have attracted much attention due to their unique luminescent properties. The ability to convert multiple photons of lower energy to ones with higher energy through an upconversion (UC) process offers a wide range of applications for UCNPs. The emission intensities and wavelengths of UCNPs are important performance characteristics, which determine the appropriate applications. However, insufficient intensities still limit the use of UCNPs; especially the efficient emission of blue and ultraviolet (UV) light via upconversion remains challenging, as these events require three or more near-infrared (NIR) photons. The aim of the study was to enhance the blue and UV upconversion emission intensities of Tm3+ doped NaYF4 nanoparticles and to demonstrate their utility in in vitro diagnostics. As the distance between the sensitizer and the activator significantly affect the energy transfer efficiency, different strategies were explored to change the local symmetry around the doped lanthanides. One important strategy is the intentional co-doping of active (participate in energy transfer) or passive (do not participate in energy transfer) impurities into the host matrix. The roles of doped passive impurities (K+ and Sc3+) in enhancing the blue and UV upconversions, as well as in influencing the intense UV upconversion emission through excess sensitization (active impurity) were studied. Additionally, the effects of both active and passive impurity doping on the morphological and optical performance of UCNPs were investigated. The applicability of UV emitting UCNPs as an internal light source for glucose sensing in a dry chemistry test strip was demonstrated. The measurements were in agreement with the traditional method based on reflectance measurements using an external UV light source. The use of UCNPs in the glucose test strip offers an alternative detection method with advantages such as control signals for minimizing errors and high penetration of the NIR excitation through the blood sample, which gives more freedom for designing the optical setup. In bioimaging, the excitation of the UCNPs in the transparent IR region of the tissue permits measurements, which are free of background fluorescence and have a high signal-to-background ratio. In addition, the narrow emission bandwidth of the UCNPs enables multiplexed detections. An array-in-well immunoassay was developed using two different UC emission colours. The differentiation between different viral infections and the classification of antibody responses were achieved based on both the position and colour of the signal. The study demonstrates the potential of spectral and spatial multiplexing in the imaging based array-in-well assays.

Visual Customer Segmentation and Data Exploration with the Self-Organizing Map

Advancements in information technology have made it possible for organizations to gather and store vast amounts of data of their customers. Information stored in databases can be highly valuable for organizations. However, analyzing large databases has proven to be difficult in practice. For companies in the retail industry, customer intelligence can be used to identify profitable customers, their characteristics, and behavior. By clustering customers into homogeneous groups, companies can more effectively manage their customer base and target profitable customer segments. This thesis will study the use of the self-organizing map (SOM) as a method for analyzing large customer datasets, clustering customers, and discovering information about customer behavior. Aim of the thesis is to find out whether the SOM could be a practical tool for retail companies to analyze their customer data.