Browsing by study line "no specialization"

Chemical reaction optimization is an iterative process that targets identifying reaction conditions that maximize reaction output, typically yield. The evolution of optimization techniques has progressed from intuitive approaches to simple heuristics, and more recently, to statistical methods such as Design of Experiments approach. Bayesian optimization, which iteratively updates beliefs about a response surface and suggests parameters both exploiting conditions near the known optima and exploring uncharted regions, has shown promising results by reducing the number of experiments needed for finding the optimum in various optimization tasks. In chemical reaction optimization, the method allows minimizing the number of experiments required for finding the optimal reaction conditions. Automated tools like pipetting robots hold potential to accelerate optimization by executing multiple reactions concurrently. The integration of Bayesian optimization to automation reduces not only the workload and throughput but also optimization efficiency. However, adoption of these advanced techniques faces a barrier, as chemists often lack proficiency in machine learning and programming. To bridge this gap, Automated Chemical Reaction Optimization Software (ACROS) is introduced. This tool orchestrates an optimization loop: Bayesian optimization suggests reaction candidates, the parameters are translated into commands for a pipetting robot, the robot executes the operations, a chemist interprets the results, and data is fed back to the software for suggesting the next reaction candidates. The optimization tool was evaluated empirically using a numerical test function, in a Direct Arylation reaction dataset, and in real-time optimization of Sonogashira and Suzuki coupling reactions. The findings demonstrate that Bayesian optimization efficiently identifies optimal conditions, outperforming Design of Experiments approach, particularly in optimizing discrete parameters in batch settings. Three acquisition functions; Expected Improvement, Log Expected Improvement and Upper Confidence Bound; were compared. It can be concluded that expected improvement-based methods are more robust, especially in batch settings with process constraints.

Positron annihilation lifetime spectroscopy (PALS) is a method used to analyse the properties of materials, namely their composition and what kind of defects they might consist of. PALS is based on the annihilation of positrons with the electrons of a studied material. The average lifetime of a positron coming into contact with a studied material depends on the density of electrons in the surroundings of the positron, with higher densities of electrons naturally resulting in faster annihilations on average. Introducing positrons in a material and recording the annihilation times results in a spectrum that is, in general, a noisy sum of exponential decays. These decay components have lifetimes that depend on the different density areas present in the material, and relative intensities that depend on the fractions of each area in the material. Thus, the problem in PALS is inverting the spectrum to get the lifetimes and intensities, a problem known as exponential analysis in general. A convolutional neural network architecture was trained and tested on simulated PALS spectra. The aim was to test whether simulated data could be used to train a network that could predict the components of PALS spectra accurately enough to be usable on spectra gathered from real experiments. Reasons for testing the approach included trying to make the analysis of PALS spectra more automated and decreasing user-induced bias compared to some other approaches. Additionally, the approach was designed to require few computational resources, ideally being trainable and usable on a single computer. Overall, testing showed that the approach has some potential, but the prediction performance of the network depends on the parameters of the components of the target spectra, with likely issues being similar to those reported in previous literature. In turn, the approach was shown to be sufficiently automatable, particularly once training has been performed. Further, while some bias is introduced in specifying the variation of the training data used, this bias is not substantial. Finally, the network can be trained without considerable computational requirements within a sensible time frame.

Fault management in mobile networks is required for detecting, analysing, and fixing problems appearing in the mobile network. When a large problem appears in the mobile network, multiple alarms are generated from the network elements. Traditionally Network Operations Center (NOC) process the reported failures, create trouble tickets for problems, and perform a root cause analysis. However, alarms do not reveal the root cause of the failure, and the correlation of alarms is often complicated to determine. If the network operator can correlate alarms and manage clustered groups of alarms instead of separate ones, it saves costs, preserves the availability of the mobile network, and improves the quality of service. Operators may have several electricity providers and the network topology is not correlated with the electricity topology. Additionally, network sites and other network elements are not evenly distributed across the network. Hence, we investigate the suitability of a density-based clustering methods to detect mass outages and perform alarm correlation to reduce the amount of created trouble tickets. This thesis focuses on assisting the root cause analysis and detecting correlated power and transmission failures in the mobile network. We implement a Mass Outage Detection Service and form a custom density-based algorithm. Our service performs alarm correlation and creates clusters of possible power and transmission mass outage alarms. We have filed a patent application based on the work done in this thesis. Our results show that we are able to detect mass outages in real time from the data streams. The results also show that detected clusters reduce the number of created trouble tickets and help reduce of the costs of running the network. The number of trouble tickets decreases by 4.7-9.3% for the alarms we process in the service in the tested networks. When we consider only alarms included in the mass outage groups, the reduction is over 75%. Therefore continuing to use, test, and develop implemented Mass Outage Detection Service is beneficial for operators and automated NOC.

Tailoring a hybrid surface or any complex material to have functional properties that meet the needs of an advanced device or drug requires knowledge and control of the atomic level structure of the material. The atomistic configuration can often be the decisive factor in whether the device works as intended, because the materials' macroscopic properties - such as electrical and thermal conductivity - stem from the atomic level. However, such systems are difficult to study experimentally and have so far been infeasible to study computationally due to costly simulations. I describe the theory and practical implementation of a 'building block'-based Bayesian Optimization Structure Search (BOSS) method to efficiently address heterogeneous interface optimization problems. This machine learning method is based on accelerating the identification of a material's energy landscape with respect to the number of quantum mechanical (QM) simulations executed. The acceleration is realized by applying likelihood-free Bayesian inference scheme to evolve a Gaussian process (GP) surrogate model of the target landscape. During this active learning, various atomic configurations are iteratively sampled by running static QM simulations. An approximation of using chemical building blocks reduces the search phase space to manageable dimensions. This way the most favored structures can be located with as little computation as possible. Thus it is feasible to do structure search with large simulation cells, while still maintaining high chemical accuracy. The BOSS method was implemented as a python code called aalto-boss between 2016-2019, where I was the main author in co-operation with Milica Todorović and Patrick Rinke. I conducted a dimensional scaling study using analytic functions, which quantified the scaling of BOSS efficiency for fundamentally different functions when dimension increases. The results revealed the target function's derivative's important role to the optimization efficiency. The outcome will help people with choosing the simulation variables so that they are efficient to optimize, as well as help them estimate roughly how many BOSS iterations are potentially needed until convergence. The predictive efficiency and accuracy of BOSS was showcased in the conformer search of the alanine dipeptide molecule. The two most stable conformers and the characteristic 2D potential energy map was found with greatly reduced effort compared to alternative methods. The value of BOSS in novel materials research was showcased in the surface adsorption study of bifenyldicarboxylic acid on CoO thin film using DFT simulations. We found two adsorption configurations which had a lower energy than previous calculations and approximately supported the experimental data on the system. The three applications showed that BOSS can significantly reduce the computational load of atomistic structure search while maintaining predictive accuracy. It allows material scientists to study novel materials more efficiently, and thus help tailor the materials' properties to better suit the needs of modern devices.

Acute lymphoblastic leukemia (ALL) is a hematological malignancy that is characterized by uncontrolled proliferation and blocked maturation of lymphoid progenitor cells. It is divided into B- and T-cell types both of which have multiple subtypes defined by different somatic genetic changes. Also, germline predisposition has been found to play an important role in multiple hematological malignancies and several germline variants that contribute to the ALL risk have already been identified in pediatric and familial settings. There are only few studies including adult ALL patients but thanks to the findings in acute myeloid leukemia, where they found the germline predisposition to consider also adult patients, there is now more interest in studying adult patients. The prognosis of adult ALL patients is much worse compared to pediatric patients and many are still lacking clear genetic markers for diagnosis. Thus, identifying genetic lesions affecting ALL development is important in order to improve treatments and prognosis. Germline studies can provide additional insight on the predisposition and development of ALL when there are no clear somatic biomarkers. Single nucleotide variants are usually of interest when identifying biomarkers from the genome, but also structural variants can be studied. Their coverage on the genome is higher than that of single nucleotide variants which makes them suitable candidates to explore association with prognosis. Copy number changes can be detected from next generation sequencing data although the detection specificity and sensitivity vary a lot between different software. Current approach is to identify the most likely regions with copy number change by using multiple tools and to later validate the findings experimentally. In this thesis the copy number changes in germline samples of 41 adult ALL patients were analyzed using ExomeDepth, CODEX2 and CNVkit.

Long term monitoring programs gather important data to understand population trends and man- age biodiversity, including phenological data. The sampling of such data can suffer from left- censoring where the first occurrence of an event coincides with the first sampling time. This can lead to overestimation of the timing of species’ life history events and obscure phenological trends. This study develops a Bayesian survival model to predict and impute the true first occurrence times of Finnish moths in a given sampling season in left-censored cases, thereby estimating the amount of left-censoring and effectively "decensoring" the data. A simulation study was done to test the model on synthetic data and explore how effect size, the severity of censoring, and sampling fre- quency effect the inference. Forward feature selection was done over environmental covariates for a generalized linear survival model with logit link, incorporating both left-censoring and interval censoring. Five-fold cross validation was done to select the best model and see what covariates would be added during the feature selection process. The validation tested the model both in its ability to predict points that were not left-censored and those that were artificially left-censored. The final model included terms for cumulative Growing Degree Days, cumulative Chilling Days, mean spring temperature, cumulative rainfall, and daily minimum temperature, in addition to an intercept term. It was trained on all of the data and predictions were made for the true first occurrence times of the left-censored sites and years.

Despite development in many areas of machine learning in recent decades, still, changing data sources between the domain in a model is trained and the domain in the same model is used for predictions is a fundamental and common problem. In the area of domain adaptation, these circum- stances have been studied by incorporating causal knowledge about the information flow between features to be utilized in the feature selection for the model. That work has shown promising results to accomplish so-called invariant causal prediction, which means a prediction performance is immune to the change levels between domains. Within these approaches, recognizing the Markov blanket to the target variable has served as a principal workhorse to find the optimal starting point. In this thesis, we continue to investigate closely the property of invariant prediction performance within Markov blankets to target variable. Also, some scenarios with latent parents involved in the Markov blanket are included to understand the role of the related covariates around the latent parent effect to the invariant prediction properties. Before the experiments, we cover the concepts of Makov blankets, structural causal models, causal feature selection, covariate shift, and target shift. We also look into ways to measure bias between changing domains by introducing transfer bias and incomplete information bias, as these biases play an important role in the feature selection, often being a trade-off situation between these biases. In the experiments, simulated data sets are generated from structural causal models to conduct the testing scenarios with the changing conditions of interest. With different scenarios, we investigate changes in the features of Markov blankets between training and prediction domains. Some scenarios involve changes in latent covariates as well. As result, we show that parent features are generally steady predictors enabling invariant prediction. An exception is a changing target, which basically requires more information about the changes in other earlier domains to enable invariant prediction. Also, emerging with latent parents, it is important to have some real direct causes in the feature sets to achieve invariant prediction performance.

Visual simultaneous localization and mapping (visual SLAM) is a method for consistent self-contained localization using visual observations. Visual SLAM can produce very precise pose estimates without any specialized hardware, enabling applications such as AR navigation. The use of visual SLAM in very large areas and over long distances is not presently possible due to a number of significant scalability issues. In this thesis, these issues are discussed and solutions for them explored, culminating in a concept for a real-time city-scale visual SLAM system. A number of avenues for future work towards a practical implementation are also described.

Understanding customer behavior is one of the key elements in any thriving business. Dividing customers into different groups based on their distinct characteristics can help significantly when designing the service. Understanding the unique needs of customer groups is also the basis for modern marketing. The aim of this study is to explore what types of customer groups exist in an entertainment service business. In this study, customer segmentation is conducted with k-prototypes, a variation of k-means clustering. K-prototypes is a machine learning approach partitioning a group of observations into subgroups. These subgroups have little variation within the group and clear differences when compared to other subgroups. The advantage of k-prototypes is that it can process both categorical and numeric data efficiently. The results show that there are significant and meaningful differences between customer groups emerging from k-prototypes clustering. These customer groups can be targeted based on their unique characteristics and their reactions to different types of marketing actions vary. The unique characteristics of the customer groups can be utilized to target marketing actions better. Other possibilities to benefit from customer segmentation include such as personalized views, recommendations and helping strategy level decision making when designing the service. Many of these require further technical development or deeper understanding of the segments. Data selection as well as the quality of the data has an impact on the results and those should be considered carefully when deciding future actions on customer segmentation.

Many researchers use dwell time, a measure of engagement with information, as a universal measure of importance in judging the relevance and usefulness of information. However, it may not fully account for individual cognitive variations. This study investigates how individual differences in cognitive abilities, specifically working memory and processing speed, can significantly impact dwell time. We examined the browsing behavior of 20 individuals engaged in information-intensive tasks, measuring their cognitive abilities, tracking their web page visits, and asses their perceived relevance and usefulness of the information encountered. Our findings shows a clear connection between working memory, processing speed, and dwell time. Based on this finding, we developed a model that combines both cognitive abilities and dwell time to predict the relevance and usefulness of web pages. Interestingly, our analysis reveals that cognitive abilities, particularly fluid intelligence, significantly influence dwell time. This highlights the importance of incorporating individual cognitive differences into prediction models to improve their accuracy. Thus, personalized services that set dwell time thresholds based on individual users' cognitive abilities could provide more accurate estimations of what users find relevant and useful.