Det finns ett behov av användarvänliga, anpassningsbara och tillgängliga journalsystem inom en variation av branscher. Befintliga system har visat sig vara bristfälliga inom dessa tre aspekter. Bristerna bottnar i att systemen sällan är anpassade för det specifika ändamålet, vilket i många fall grundar sig i att slutanvändarna inte varit med i utvecklingsprocessen. När användare har implementerat befintliga journalsystem inom sina verksamheter har även bristfälligheter gällande tillgänglighet påvisats. På den utvecklade journalsystemprototypen som arbetet resulterade i, utfördes ett antal tester av uppdragsgivarens anställda. Dessa tester påvisade att prototypens användarvänlighet, anpassningsbarhet och tillgänglighet uppnådde de krav som företaget hade på ett journalsystem. Prototypen har, trots det positiva resultatet, fortfarande utvecklingspotential och områden att förbättra. / There is a need for user-friendly, customizable and accessible record systems in a variety of industries. Existing systems have proven to be deficient in these three aspects. The shortcomings are due to the fact that the systems are seldom adapted for the specific purpose, which in many cases is a consequence of the end users not being involved in the development process. When users have implemented existing journal systems within their operations, deficiencies regarding accessibility have also been identified. On the developed record system prototype that the work resulted in, a number of tests were performed by the client's employees. These tests showed that the user-friendliness, adaptability and availability of the prototype met the requirements of the company for a record system. Despite the positive result, the prototype still has development potential and areas for improvement.

Online supervised learning from fast-evolving imbalanced data streams has applications in many areas. That is, the development of techniques that are able to handle highly skewed class distributions (or 'class imbalance') is an important area of research in domains such as manufacturing, the environment, and health. Solutions should be able to analyze large repositories in near real-time and provide accurate models to describe rare classes that may appear infrequently or in bursts while continuously accommodating new instances. Although numerous online learning methods have been proposed to handle binary class imbalance, solutions suitable for multi-class streams with varying degrees of imbalance in evolving streams have received limited attention. To address this knowledge gap, the first contribution of this thesis introduces the Online Learning from Imbalanced Multi-Class Streams through Dynamic Sampling (DynaQ) algorithm for learning in such multi-class imbalanced settings. Our approach utilizes a queue-based learning method that dynamically creates an instance queue for each class. The number of instances is balanced by maintaining a queue threshold and removing older samples during training. In addition, new and rare classes are dynamically added to the training process as they appear. Our experimental results confirm a noticeable improvement in minority-class detection and classification performance. A comparative evaluation shows that the DynaQ algorithm outperforms the state-of-the-art approaches. Our second contribution in this thesis focuses on fairness-aware learning from imbalanced streams. Our work is motivated by the observation that the decisions made by online learning algorithms may negatively impact individuals or communities. Indeed, the development of approaches to handle these concerns is an active area of research in the machine learning community. However, most existing methods process the data in offline settings and are not directly suitable for online learning from evolving data streams. Further, these techniques fail to take the effects of class imbalance, on fairness-aware supervised learning into account. In addition, recent fairness-aware online learning supervised learning approaches focus on one sensitive attribute only, which may lead to subgroup discrimination. In a fair classification, the equality of fairness metrics across multiple overlapping groups must be considered simultaneously. In our second contribution, we thus address the combined problem of fairness-aware online learning from imbalanced evolving streams, while considering multiple sensitive attributes. To this end, we introduce the Multi-Sensitive Queue-based Online Fair Learning (MQ-OFL) algorithm, an online fairness-aware approach, which maintains valid and fair models over evolving streams. MQ-OFL changes the training distribution in an online fashion based on both stream imbalance and discriminatory behavior of the model evaluated over the historical stream. We compare our MQ-OFL method with state-of-art studies on real-world datasets and present comparative insights on the performance. Our final contribution focuses on explainability and interpretability in fairness-aware online learning. This research is guided by the concerns raised due to the black-box nature of models, concealing internal logic from users. This lack of transparency poses practical and ethical challenges, particularly when these algorithms make decisions in finance, healthcare, and marketing domains. These systems may introduce biases and prejudices during the learning phase by utilizing complex machine learning algorithms and sensitive data. Consequently, decision models trained on such data may make unfair decisions and it is important to realize such issues before deploying the models. To address this issue, we introduce techniques for interpreting the outcomes of fairness-aware online learning. Through a case study predicting income based on features such as ethnicity, biological sex, age, and education level, we demonstrate how our fairness-aware learning process (MQ-OFL) maintains a balance between accuracy and discrimination trade-off using global and local surrogate models.

No description available.

Nanopositioners have been developed with electrostatic, piezoelectric, magnetic, thermal, and electrochemical actuators. They move with as many as six degrees of freedom; some are composed of multiple stages that stack together. Both macro-scale and micro-scale nanopositioners have been fabricated. A summary of recent research in micropositioning and nanopositioning is presented to set the background for this work. This research project demonstrates that a dual-stage nanopositioner can be created with microelectromechanical systems technology such that the two stages are integrated on a single silicon chip. A nanopositioner is presented that has two stages, one for coarse motion and one for fine motion; both are fabricated by surface micromachining. The nanopositioner has one translational degree of freedom. Thermal microactuators operate both stages. The first stage includes a bistable mechanism: it travels 52 micrometers between two discrete positions. The second stage is mounted on the first stage and moves continuously through an additional 8 micrometers in the same direction as the first stage. Two approaches to the control of the second stage are evaluated: first, an electrical input is transmitted to an actuator that moves with the first stage; second, a mechanical input is applied to an amplifier mechanism mounted on the first stage after completing the coarse motion. Four devices were designed and fabricated to test these approaches; the one that performed best was selected to fulfill the objective of this work. Thermal analysis of the actuators was performed with previously developed tools. Pseudo-rigid-body models and finite element models were created to analyze the mechanical behavior of the devices. The nanopositioners were surface micromachined in a two-layer polysilicon process. Experiments were performed to characterize the resolution, repeatability, hysteresis, and drift of the second stages of the nanopositioners with open-loop control. Position measurements were obtained from scanning electron micrographs by a numerical procedure, which is described in detail. The selected nanopositioner demonstrated 170-nanometer resolution and repeatability within 37 nanometers. The hysteresis of the second stage was 6% of its full range. The nanopositioner drifted 25 nanometers in the first 60 minutes of operation with a time constant of about 6 minutes. The dual-stage nanopositioner may be useful for applications such as variable optical attenuators or wavelength-specific add--drop devices.

Background: Rising global uncertainty and volatility have changed how businesses envision themselves and their future. Particularly Social Business Hybrids emphasize the importance of purpose beyond profit. Their aim is to develop a more inclusive and green economy by pursuing and creating both financial and social value. This aim, however, comes with the risk of mission drift, meaning that the organization could prioritize one value at the expense of the other. As this is an emergent strategic dilemma in the field of social entrepreneurship, it is prudent to find ways to manage the risk of mission drift. In this context, considerable attention has to be given to the question of how employees of Social Business Hybrids manage the risk of mission drift, as their viewpoint has mostly been neglected by prior scholarship. Purpose: The purpose of the present study is to provide an understanding of why the management of mission drift is relevant for Social Business Hybrids and to give new insights into perspectives that employees of Social Business Hybrids have towards tactics and strategies that support successful management of mission drift. By researching the critical role that employees play as stakeholders in the context of mission drift management, we aim to enrich current literature by deriving new insights into strategies that can help Social Business Hybrids successfully balance their dual objectives. Method: This study is based on: Qualitative, inductive research; Ontology – Relativism; Epistemology – Social Constructionism; Methodology – Grounded Theory; Data Collection – 12 semi-structured in-depth Interviews; Sampling – Purposive, Snowball; Data Analysis – Grounded Analysis  Conclusion: In our findings, we ascertained distinct sources that anticipate a risk for mission drift. Further, we identified detailed tactics that can significantly support the management of mission drift. Finally, resulting from the findings, a framework was developed, that proposes five distinctive overarching strategies, which are enabled by two underlying mechanisms that employees in Social Business Hybrids apply to manage mission drift.

I detta examensarbete behandlas en jämförelse mellan två olika material utifrån LCC för GC-broar. Dessa är stål och Delignified Densified Wood (DDW). Samhället är i behov av att utvecklas inom ett mer hållbart- och anpassad till klimatkraven. Detta innebär att det ställs krav på Trafikverket. Dessa krav innebär bland annat att byggnadsmaterial behöver utvecklas. Sverige har god tillgång till skog och träframtagning. Trafikverket bör nyttja mer av träets potential. Denna studie har därför gjorts för att bedöma DDW som är ett träbaserat material utifrån intervjuer och livscykelkostnadsanalys (LCC-analys). LCC är ett tillvägagångsätt för att få en totalbild över en produkts samtliga kostnader under dess livslängd. Det vill säga att för en bro bedöms kostnaderna från projektering till rivning. Syftet med LCC-analys är att hitta den mest lönsamma investeringskostnad som uppfyller dagens samhällskrav. DDW är framtagen av Forskare vid Marylands Universitet. Forskningen har observerat att obehandlat trä går att få lika starkt som stål. Processen innebär att ligninet avlägsnas och träet placeras i tryck under värme på cirka 100 C°. Detta leder till att cellulosafibrerna pressas samman och blir hårdare. Det resulterar förenklat sagt att träet blir mycket hårdare, tåligare och starkare och går att forma och böja. Resultatet av examensarbetet tyder på att DDW inte är lämpligt just nu som konstruktionsmaterial i utomhusklimat och inte är ekonomiskt lönsamt. DDW är vetenskapligt intressant och bör forskas vidare. I dagsläget rekommenderas inte DDW som GC-bromaterial. Trafikverket bör bygga mer GC- broar av trä som passar väl som GC-bromaterial. / n this thesis a comparison is made between two different construction materials based on LCC for pedestrian and cycle bridges. These materials are steel and Delignified Densified Wood (DDW). Today ́s society needs to develop with more sustainable and climate-friendly construction. This need imposes demands on the Swedish transport administration, e.g. that building materials need to be developed. One viable material for Sweden, since it has good access to forestry and wood production, is wood. The Swedish transport administration should use more of the potential of wood. This study therefore assesses DDW, which is a wood-based material, based on interviews and life cycle cost analysis (LCC-analysis). LCC is a way of getting a complete picture of a product's entire costs during its lifespan. In the case of a bridge, the costs are estimated from planning to demolition. The purpose of LCC analysis is to find the most profitable investment cost that meets today's social requirements. DDW is developed by researchers at Maryland University. Researchers have observed that natural wood can be as strong as steel. The process involves removing lignin and placing it under pressure at a temperature of about 100 ° C. This causes the cellulosic fibers to compress and become harder. This simply means that the wood becomes much harder, more durable and stronger, and can also shape and bend. The result of the thesis suggests that DDW is not suitable at present as construction material in outdoor climate and is not economically profitable. DDW is scientifically interesting and should be researched further.

Steel plate shear walls (SPSWs) are a lateral force resisting system consisting of thin infill steel plates surrounded by boundary frame members. The infill steel plates are allowed to buckle in shear and subsequently form diagonal tension field actions during earthquake events. Hysteretic energy dissipation of this system is primarily achieved through yielding of the infill plates. Conceptually, in a SPSW system with ideally rigid columns pinned to ground, the infill plates at different stories will yield simultaneously as a result of the lateral loads. However, when the columns become flexible, infill plate yielding may initially occur at one story and progressively spread into the other stories with increasing roof displacement. This research investigates the effect of column stiffness on infill plate yielding sequence and distribution along the height of steel plate shear walls subjected to earthquake forces. Analytical models are derived and validated for two-story SPSWs. Based on the derived model, probabilistic simulations are conducted to calculate the probability of achieving infill plate yielding in both stories before occurrence of a premature failure caused by excessive inter story drift at the initially yielded story. A total of three simulation methods including the Monte-Carlo method, the Latin Hypercube sampling method, and the Rosenblueth’s 2K+1 point estimate method were considered to account for the uncertain infill plate thickness and lateral force distributions in the system.The investigation is also extended to multi-story SPSWs. Three example six-story SPSWs are evaluated using the Rosenblueth's 2K+1 point estimation method which is identified to be most efficient from the simulation on two-story SPSWs. Moreover, the effectiveness of the column minimum moment of inertia required in the current code for achieving infill plate yielding at every story of SPSWs is evaluated.

No / A vehicle that deviates laterally from its intended path of travel when the brakes are applied is considered to demonstrate ‘instability’ in the form of an unexpected and undesirable response to the driver input. Even where the magnitude of lateral displacement of the vehicle is small (i.e. ‘drift’ rather than ‘pull’) such a condition would be considered unacceptable by manufacturers and customers. Steering ‘drift’ during braking can be caused by several factors, some of which relate to vehicle design and others to external influences such as road conditions. The study presented here examines the causes and effects of steering drift during straight-line braking. A comparative analysis is made between two types of vehicle model: one built with rigid suspension components and the other with flexible components. In both the cases, the vehicle behaviour is simulated during braking in a straight line, and responses including lateral acceleration, yaw rate, and lateral displacement of the vehicle are predicted and analysed under fixed steering control. Suspension/steering geometry characteristics, namely toe steer and caster angle, have been studied to understand how the effect of variations in these parameters differs in models with rigid or flexible components drift during straight-line braking. Results from both vehicle models show that differences between rigid and flexible components can affect the predicted steering drift propensity. The differences between the two models have emphasized the importance of using flexible (compliant) components in vehicle handling simulations to achieve better correlation between prediction and experiment.

Post Earnings Announcement Drift (PEAD) is a market anomaly that challenge the “Efficient Market Hypothesis” (EMH). It was first discovered in 1968 by Ball and Brown. When firms on the stock market have their earnings announcement the stock price will be affected and tend to drift up or down in price for days, weeks or months. Based on the limited research studies available there is acceptance that PEAD exists in the Stockholm stock exchange but depending on how measured the effect can strongly differ. In this master thesis we will study PEAD anomaly in the Swedish stock market and how pronounced it is on the stock’s sector, beta and trading volume. This study is an event and quantitative study which analyses firms on the Stockholm exchange market during the period between January 2007 to December 2022. A price measurement methodology has been used where the benchmark for abnormal (or excess) returns is the index of the list. Evidence shows that PEAD is present in the Stockholm Stock Exchange but that the effect is limited. The fact that the event abnormal returns are significant regarding of the returns up to after 60 trading days (although on a very small effect) provides insight and understanding of the effect. This study has also provided insight that beta and sector is a relevant PEAD parameter, maybe as important as the abnormal returns in the event itself. Trading volume have not provided any insight on PEAD in this study.

Time series anomaly detection receives increasing research interest given the growing number of data-rich application domains. Recent additions to anomaly detection methods in research literature include deep learning algorithms. The nature and performance of these algorithms in sequence analysis enable them to learn hierarchical discriminating features and time-series temporal nature. However, their performance is affected by the speed at which the time series arrives, the use of a fixed threshold, and the assumption of Gaussian distribution on the prediction error to identify anomalous values. An exact parametric distribution is often not directly relevant in many applications and it’s often difficult to select an appropriate threshold that will differentiate anomalies with noise. Thus, implementations need the Prediction Interval (PI) that quantifies the level of uncertainty associated with the Deep Neural Network (DNN) point forecasts, which helps in making a better-informed decision and mitigates against false anomaly alerts. To achieve this, a new anomaly detection method is proposed that computes the uncertainty in estimates using quantile regression and used the quantile interval to identify anomalies. Similarly, to handle the speed at which the data arrives, an online anomaly detection method is proposed where a model is trained incrementally to adapt to the concept drift that improves prediction. This is implemented using a window-based strategy, in which a time series is broken into sliding windows of sub-sequences as input to the model. To adapt to concept drift, the model is updated when changes occur in the new arrival instances. This is achieved by using anomaly likelihood which is computed using the Q-function to define the abnormal degree of the current data point based on the previous data points. Specifically, when concept drift occurs, the proposed method will mark the current data point as anomalous. However, when the abnormal behavior continues for a longer period of time, the abnormal degree of the current data point will be low compared to the previous data points using the likelihood. As such, the current data point is added to the previous data to retrain the model which will allow the model to learn the new characteristics of the data and hence adapt to the concept changes thereby redefining the abnormal behavior. The proposed method also incorporates feature extraction to capture structural patterns in the time series. This is especially significant for multivariate time-series data, for which there is a need to capture the complex temporal dependencies that may exist between the variables. In summary, this thesis contributes to the theory, design, and development of algorithms and models for the detection of anomalies in both static and evolving time series data. Several experiments were conducted, and the results obtained indicate the significance of this research on offline and online anomaly detection in both static and evolving time-series data. In chapter 3, the newly proposed method (Deep Quantile Regression Anomaly Detection Method) is evaluated and compared with six other prediction-based anomaly detection methods that assume a normal distribution of prediction or reconstruction error for the identification of anomalies. Results in the first part of the experiment indicate that DQR-AD obtained relatively better precision than all other methods which demonstrates the capability of the method in detecting a higher number of anomalous points with low false positive rates. Also, the results show that DQR-AD is approximately 2 – 3 times better than the DeepAnT which performs better than all the remaining methods on all domains in the NAB dataset. In the second part of the experiment, sMAP dataset is used with 4-dimensional features to demonstrate the method on multivariate time-series data. Experimental result shows DQR-AD have 10% better performance than AE on three datasets (SMAP1, SMAP3, and SMAP5) and equal performance on the remaining two datasets. In chapter 5, two levels of experiments were conducted basis of false-positive rate and concept drift adaptation. In the first level of the experiment, the result shows that online DQR-AD is 18% better than both DQR-AD and VAE-LSTM on five NAB datasets. Similarly, results in the second level of the experiment show that the online DQR-AD method has better performance than five counterpart methods with a relatively 10% margin on six out of the seven NAB datasets. This result demonstrates how concept drift adaptation strategies adopted in the proposed online DQR-AD improve the performance of anomaly detection in time series. / Petroleum Technology Development Fund (PTDF)