A deeper understanding of Multiple Sclerosis (MS) symptom progression is required for diagnostic accuracy and patient care. Remote monitoring through smartphones can provide continuous insights in the well-being of MS patients. This research aims to explore differences between MS and healthy typing behavior through collected typing behavior data (keystroke dynamics) on smartphones. The main contribution of this work was finding patterns within typing data that distinguish healthy typing behavior from MS with unlabeled time series data, except the diagnosis. A dataset of three combined studies was provided, which included 182 patients suffering from MS and 335 healthy subjects. An autoencoder was fitted on healthy samples to learn the temporal patterns in healthy typing behavior. Then, an anomaly detection method leveraged both the reconstruction loss and embeddings of these samples to determine whether MS samples deviated from healthy behavior. MS subjects were evaluated against healthy subjects through the ROC-AUC on both sample level and subject level. An initial hold out dataset of 20\% of the subjects was left out for validation. The overall performance was poor (AUC 0.6-0.8), because we aimed to find one general method for all MS subjects. Key limitations were the large variance in healthy typing behavior, relatively short window lengths, and difficulty in quantification of the training step of the autoencoder. Despite the difficulty of evaluating the performance of our suggested method, the described methods provide interesting monitoring possibilities for neurologists to enhance MS diagnosis. These monitoring possibilities could extend towards monitoring the effect of treatment, which is the first step towards personalised healthcare in MS.

The intensive care unit (ICU) is a hospital department where critically ill patients requiring organ support or intensive monitoring are admitted. Nowadays, the care provided in an intensive care unit has advanced so that more patients are being discharged alive. Advances in ICU care have increased patient survival rates, yet ICU admission is still associated with high morbidity and mortality both during and after the stay.

Identifying patients at high risk of complications during ICU admission is crucial. Even though there has been an increasing focus on predictive models, making accurate predictions with the data currently available remains challenging. A non-conventional parameter that appears to have promising value is heart rate variability (HRV), which reflects the fluctuations in time intervals between consecutive heartbeats and can be derived from the electrocardiogram. We investigated whether heart rate variability was able to predict ICU mortality and ICU length of stay. We employed a machine learning approach, assessing three models for each outcome. We used nested cross validation to estimate the performance and optimize and select the final models. Data from 468 adult patients admitted to the ICU for 48 hours or longer were analyzed and nine HRV measures were calculated. Two HRV measures, the power in the high frequency band and the standard deviation of 5-minute average RR intervals, showed a significant difference between the ICU survivors (n=398) and ICU non-survivors (n=71). While individual HRV measures had limited predictive power for ICU mortality, combining HRV with clinical features improved performance. The best performing model was an eXtreme Boosting Gradient classifier that used clinical features in combination with three HRV measures (power in the high frequency band, power in the low frequency band and the ratio between the power in the low frequency and high frequency band) achieving an AUC of 0.76. The models predicting ICU length of stay performed poorly, with the best model achieving a mean absolute error of 5.07 days. These findings suggest that HRV, when combined with clinical features, has potential in predicting ICU mortality, though further research is necessary before clinical implementation.

Monitoring ICU survivors after discharge is equally important, as half of these survivors suffer from Post Intensive Care Syndrome, which negatively impacts their quality of life and increases their healthcare needs. A team of researchers from the ICU conducted a pilot study establishing the feasibility of monitoring fifteen ICU survivors using eHealth, but a larger clinical trial was needed to assess feasibility on a larger scale. Therefore, we aimed to prepare for the ICU Recover Box 2.0 study. We evaluated the results and challenges of the first pilot. The study protocol was revised to include more participants and the use of new smart technology, specifically the Corsano CardioWatch, replacing the non-CE marked devices from the first pilot, which required 24/7 researcher availability. Preparations included finalizing the application to the Medical Ethical Review Committee and thinking out the study logistics, resulting in a comprehensive plan for the conduct of the study. Key takeaways from this process include the recognition that research is an iterative process of
continuous learning, that the purpose of the study must be carefully considered with ethical concerns in mind, and that the increasing importance of data requires careful planning for its security, processing and storage. The METC application has been submitted. Once approved, the study can proceed on a well-prepared basis.

In reinforcement learning, the ability to generalize to unseen situations is pivotal to an agent’s success. In this thesis, two novel methods that aim to enhance the generalizability of an agent will be introduced. Both of the methods rely on the idea that the diversity of a replay buffer increases an agent’s ability to generalize. The first utilizes the agent’s exploration strategies to reach interesting states. The second aims to reach further using an additional goal-conditioned agent. Both methods demonstrate improved adaptability without relying on domain-specific knowledge and show promising results.

In human-AI agent interactions, providing clear visual or textual explanations for the agent's actions and decisions is crucial for ensuring successful collaboration. This research investigates whether having the visual explanations displayed only on-demand, instead of having them consistently shown as the baseline, has an impact on the human supervisor's level of confidence and satisfaction with the AI agent. Therefore, a case study of 40 participants was conducted to explore this hypothesis and the participants were divided into 2 groups, one interacting with the on-demand condition, and the other with the baseline one. Through questionnaires, the participants' capacity and moral trust in the robot, the explainable artificial intelligence satisfaction, and the disagreement rate with the robot's decisions have been collected. Demographic data was gathered from the participants to explore whether their background could impact the collaboration. This data included the participants' gender, age, education, gaming experience, risk propensity, trust propensity, and utilitarianism. The resulting statistical analyses indicated no significant differences between the baseline and the on-demand conditions concerning trust and explanation satisfaction. This suggests that the overall collaboration was not primarily impacted by the frequency of visual explanations requested on demand. Although the results implied a high satisfaction with the interaction, further studies with more diverse user groups are recommended. Overall, this research reinforces the importance of transparency in decision-making processes during the collaboration between an AI agent and a human supervisor.

The integration of robots in human-robot teams, particularly in high-stakes environments like firefighting, requires effective communication and decision-making to ensure safety and efficiency. This study explores the impact of adding contrastive explanations to feature attributions in robot explanations on human-robot teamwork during firefighting simulations. Contrastive explanations aim to improve human understanding by highlighting why a robot chose one decision over another using allocations of variables. The experiment involved 40 participants, divided into two groups, each interacting with either the baseline or contrastive version of the robot in the simulated environment. Results indicate that contrastive explanations significantly increased participants' capacity trust in the robot, though they did not significantly affect moral trust. Additionally, the results showed a lower satisfaction level with the explanations given by the robot. The disagreement rate between human decisions and robot actions was lower in the contrastive group, suggesting possible enhanced understanding and agreement with the robot's decisions. These findings underscore the potential of contrastive explanations to enhance trust and collaboration in human-robot teams, paving the way for more effective integration of robots in critical operations. Future research should focus on larger sample sizes and explore the inclusion of contrastive decisions made by the robot alongside explanations to further validate these findings.

In the pursuit of employing interpretable and performant Machine Learning models, Decision Trees has become a staple in many industries while being able to produce near-optimal results. With computational power becoming more accessible, there has been increasing progress in constructing Optimal Decision Trees. It guarantees optimal solutions with respect to different metrics within a given size limit on training data while requiring a smaller number of nodes and becoming more viable to compute on real-world data. However, non-linear metrics, which are very effective when evaluating trees on imbalanced datasets, still represent a challenge regarding runtime performance and scalability. Previous approaches generate the Pareto Front of the set of possible solutions, an expensive operation in computing the optimal tree. To address this gap, we introduce a novel merging algorithm of two Pareto Fronts using convex hulls, offering better pruning and leading to an increase in scalability. The experiments show a significant improvement in runtime of almost 10\% on bigger datasets and higher-depth trees using the F1-score metric, with the potential to be applied to other convex metrics.

Survival analysis predicts survival functions that give the probability of survival until a given time. Many applications of survival analysis involve health care, which requires interpretability of the models used to predict the survival function. Provably optimal decision trees have shown to be an interpretable alternative to so-called black box models. However, these algorithms often choose estimators that are fast, yet not necessarily most accurate. Moreover, the objective functions of optimal decision tree algorithms tend to make (possibly incorrect) assumptions about the survival function. In this paper, we tackle both problems. We implement the iterative Breslow estimator in an already existing optimal survival tree algorithm in order to iteratively improve the Nelson-Aalen estimator. This approach has great potential, as we show by using it on artificial datasets, but we do not see an improvement in accuracy on real world data. To eliminate the assumptions made by the objective function, we implement the Integrated Brier Score objective, which causes a significant improvement on training accuracy. However, we see no improvement on out-of-sample accuracy

The Algorithm Selection Problem (ASP) presents a significant challenge in numerous industries, requiring optimal solutions for complex computational problems. Traditional approaches to solving ASP often rely on complex, black-box models like random forests, which are effective but lack transparency, and they often fail to balance performance with interpretability. This paper investigates the performance-interpretability trade-off for the ASP, specifically focused on Optimal Decision Trees (ODTs) as recent innovations have made the use of ODTs more viable. We compare ODTs against 4 other tree-based models, using 11 different datasets. We show there is no apparent tradeoff between performance and interpretability for ODTs which have been trained using an instance cost-sensitive approach, as they achieve comparable performance to a Random Forest Regressor while maintaining interpretability through multiple orders of magnitude fewer leaf nodes.

Decision trees are valued for their ability to logically and transparently classify data. While heuristic methods to compute such trees are efficient, they often compromise on accuracy, prompting interest in Optimal Decision Trees (ODTs), which have the best misclassification score for a given tree size limit and training dataset. The dynamic programming (DP) approach for ODTs has shown improvements over alternatives such as mixed-integer or constraint programming. That being said, it requires further improvements to handle exponential runtime scaling with the depth of the tree and the number of features of the dataset. Leveraging modern hardware, such as multiple CPU cores, offers a promising solution to improve efficiency. This paper proposes a multithreading method for DP ODTs which we apply to STreeD specifically. We introduce a shared memory model and determine which components of the original program can be made local to the threads. We investigate whether it is more efficient to start multithreading at the root of the search tree than near its leaf nodes. We find the former to be superior, resulting in faster runtimes and less shared resource access. Empirical evaluations against the state of the art demonstrate better runtimes, particularly beneficial for large datasets. Finally, thread scaling analyses reveal substantial speed-ups, exceeding 2.5 times with four threads, highlighting our approach's effectiveness for computationally-intensive tasks.

Survival analysis is a branch of statistics concerned with studying and estimating the expected time duration until some event, such as biological death, occurs. Survival distributions are fitted based on historical data, where some instances are censored, meaning that the actual time of the event is not known precisely. Survival trees extend on the classical statistical methods developed and can capture complex non-linear relations between the variables by recursively splitting the instances by generated rules and fitting a different survival distribution in each leaf. Moreover, decision trees are desirable models due to their interpretable nature. We extend existing optimal survival tree methods by considering Cox Proportional Hazard models in each leaf node, which allows us to find more complex yet interpretable relationships than existing methods. The experiments show that our model outperforms state-of-the-art methods for creating survival trees, SurTree, OST, and CTree, especially in determining the relative risks between out-of-sample observations while generating significantly smaller trees.

Counterfactual explanations (CEs) are emerging as a crucial tool in Explainable AI (XAI) for understanding model decisions. This research investigates the impact of various factors on the quality of CEs generated for classification tasks. We explore how inter-class distance, data imbalance, balancing techniques, the presence of biased classifiers, and decision thresholds influence CE quality. To answer these research questions, we conduct experiments on various datasets, classification models and counterfactual generators. The datasets include the MNIST and GMSC dataset. The models include well-established models like MLP and Random Forest, along with the novel NeuroTree model. The generators include the method proposed by Wachter et al. and the REVISE method. We evaluate how different factors affect CE quality by performing an extensive experimental analysis. Our findings demonstrate that increasing inter-class distance degrades CE quality, particularly explanation plausibility. Data imbalance showed minimal impact, while balancing techniques yielded a slight improvement in CE plausibility, especially for the minority class. Classifiers biased towards specific subgroups resulted in lower CE quality for those subgroups. We observed limited evidence for a consistent amplification effect of decision thresholds. This research utilizes various datasets and classification models, including the novel NeuroTree model. Our findings contribute to XAI by providing insights into factors affecting CE quality and highlighting areas for future development, particularly regarding fairness and handling imbalanced data.

Optical Character Recognition (OCR) is a pivotal technology used to extract text information from images, finding wide-ranging applications in document digitization and medical records management. The integration of machine learning has ushered in an era of swift and precise OCR models. Broadly, OCR comprises two key components: detecting the bounding boxes around text instances and recognizing the characters within them. Presently, prevailing OCR models are primarily intricate two-stage systems necessitating real-time operation on remote servers. Nevertheless, end-to-end models exhibit superior performance from a data utilization perspective. There exist scenarios where offline models prove indispensable, such as in environments with restricted internet access or locales with stringent data privacy and security requirements.

This project delves into various end-to-end models, leveraging the PaddleOCR end-to-end model as a foundational reference to devise a compact OCR model tailored for edge devices. Through meticulous optimization of the backbone architecture and the introduction of diverse Feature Pyramid Network (FPN) structures within the stem network, we achieved a remarkable reduction in model size, down to 19MB. This represents a substantial advancement, constituting merely one-tenth of the original PaddleOCR end-to-end model's footprint.

By leveraging an extensive database and conducting a series of fine-tuning experiments specifically tailored for end-to-end OCR tasks involving curved text images, the model exhibits an impressive precision rate of 47.3% and an f-score of 45.3%. This achievement highlights the effectiveness of the customized loss function relative to the original model, despite its reduced size. Notably, this performance is comparable to certain end-to-end models with larger backbones. Furthermore, an Android demo has been carefully developed to demonstrate the model's capabilities on mobile devices, achieving an average processing time of 433 milliseconds per image.

Intrusion detection systems (IDSs) are essential for protecting computer systems and networks from malicious attacks. However, IDSs face challenges in dealing with dynamic and imbalanced data, as well as limited label availability. In this thesis, we propose a novel elastic gradient boosting decision tree algorithm, namely Elastic CatBoost Uncertainty (eCBU), that adapts to concept drifts and copes with label scarcity by using a novel sequential uncertainty estimation method. We compare our method with state-of-the-art techniques on synthetic and real-world datasets and show that it achieves comparable accuracy and higher robustness to limited label availability in intrusion detection tasks.

Bacterial identification is crucial for addressing infectious diseases and enabling effective treatment strategies. Conventional bacteria identification methods like MALDI-TOF, while efficient, lack the capability for screening the effectiveness of antibiotics. On the other hand, existing antimicrobial resistance (AMR) tests, despite being reliable, suffer from time inefficiency and lack concurrent identification capabilities. In response to these challenges, the present study employs the recent advancements in single-cell nanomotion detection using graphene drums to address these limitations. We integrate nanomotion detection with Machine Learning (ML) algorithms which enables us to simultaneously identify bacteria phenotype and their resistance to antibiotics. Bacterial time signals are transformed into time-frequency spectrograms, which then serve as inputs for machine learning algorithms. Through pattern recognition, these algorithms identify features within the images, facilitating the development of robust classification models. Utilizing single-cell nanomotion signals, differentiation is achieved between the species Escherichia coli, Staphylococcus aureus, and Klebsiella pneumoniae, as well as in detecting antibiotic-resistant and -susceptible strains, achieving an accuracy of 98.57% in the latter. This research marks the first instance of ML integrated with nanomotion detection for bacterial species identification and antibiotic susceptibility testing. It provides a basis for advanced diagnostic tools, expediting the provision of vital data regarding bacterial identification and antibiotic susceptibility, contributing significantly to medical diagnostics.

Federated learning is a privacy-enforcing machine learning technology but suffers from limited scalability. This limitation mostly originates from the internet connection and memory capacity of the central parameter server, and the complexity of the model aggregation function. Decentralized learning has recently been emerging as a promising alternative to federated learning. This novel technology eliminates the need for a central parameter server by decentralizing the model aggregation across all participating nodes. Numerous studies have been conducted on improving the resilience of federated learning against poisoning and Sybil attacks, whereas the resilience of decentralized learning remains largely unstudied. This research gap serves as the main motivator for this study, in which our objective is to improve the Sybil poisoning resilience of decentralized learning.

We present SybilWall, an innovative algorithm focused on increasing the resilience of decentralized learning against targeted Sybil poisoning attacks. By combining a Sybil-resistant aggregation function based on similarity between Sybils with a novel probabilistic gossiping mechanism, we establish a new benchmark for scalable, Sybil-resilient decentralized learning.

A comprehensive empirical evaluation demonstrated that SybilWall outperforms existing state-of-the-art solutions designed for federated learning scenarios and is the only algorithm to obtain consistent accuracy over a range of adversarial attack scenarios. We also found SybilWall to diminish the utility of creating many Sybils, as our evaluations demonstrate a higher success rate among adversaries employing fewer Sybils. Finally, we suggest a number of possible improvements to SybilWall and highlight promising future research directions.

This paper presents a comparative study of multiple algorithms that can be used to automatically search for high-performing pipelines on machine learning problems. These algorithms, namely Very Large-Scale Neighbourhood search (VLSN), Breadth-first search, Metropolis-Hastings, Monte-Carlo tree search (MCTS), enumerative A* search, and Genetic Programming, are evaluated on three datasets. The performance of VLSN is consistently acceptable for this task, but the best performance is given by MCTS. Interestingly the results show that limiting the solution space to pipelines containing only a classifier operator does not significantly decrease performance. Three possible explanations for this are that the datasets used are too simple, the use of default hyperparameters makes preprocessing and feature selection operators useless, or the evaluation of pipelines on a limited training set makes the search procedure ineffective. This research contributes to the field of AutoML by shedding light on algorithm performance and providing insights for future improvements.

In AutoML, the search space of possible pipelines is often large and multidimensional. This makes it very important to use an efficient search algorithm. We measure the effectiveness of the Metropolis-Hastings algorithm (M-H) in a pipeline synthesis framework, when the search space is described by a context-free grammar. We also compare the performance of the M-H algorithm to other search algorithms. While AutoML frameworks use many different search algorithms, and comparisons between AutoML frameworks exist, this is the first paper that compares the performance of different search algorithms in the context of pipeline synthesis under equal conditions. We found that M-H is slightly outperformed by BFS2, the simplest possible search algorithm. We conclude that the datasets we use for evaluating the algorithms are too simple to meaningfully compare the performance of different search algorithms. We also conclude that for simple datasets, simple search algorithms work best.

Cardiovascular diseases are one of the primary causes of mortality worldwide. Paroxysmal atrial fibrillation is a specific type that is difficult to detect and diagnose in a short time frame. To overcome this, we investigated if long-term wearable data can be used for the detection of heart diseases. The BigIdeasLab_STEP dataset and long-term Fitbit data from the ME-TIME study were used to examine this.

Our analysis showed a correlation between the window/stride size and accuracy when performing activity classification with the BigIdeasLAB_STEP dataset. Moreover, variability was found between subjects due to differences in the physical structure of their hearts. Normalization proved to be a crucial step to minimize the subject variability and significantly improved performance. Grouping subjects and performing classification inside a group helped to improve performance and decrease inter-subject variability. Integrating handcrafted features in deep learning networks also improved classification performance.

Analysis of the long-term Fitbit data showed that there is a difference between individuals based on their health condition. Classification of individual peaks was possible and worked best when utilizing a time series-specific support vector machine and grouping peaks together. Grouping peaks per week from a person and calculating a percentage of heart disease-predicted peaks also worked relatively well to distinguish between heart disease and reference subjects. Like with the previous dataset, normalization proved to be a crucial step to minimize subject variability.

The findings indicate that heart rate time series can be utilized for classification tasks like predicting activity or the detection of heart diseases. However, normalization and grouping techniques need to be chosen carefully to minimize the issue of subject variability.

In image search, an algorithm tries to identify images in a database that are similar to a query image. Image search has numerous applications. For example, image search can help historians find images of a historical building from a large image database of buildings worldwide. Feature extraction and nearest neighbors methods are standard steps in the image search task. Current state-of-the-art image search engines extract features from images through CNNs. Next, they find top-n images closest to the query image by nearest neighbor search (NN search). After NN search, the engine outputs results. However, it is still hard for current image search technologies to identify images under complicated situations, such as illumination, viewpoints, and rotation changes. These changes will reduce the accuracy of the engine. Therefore, how to improve the accuracy of the image search engine is still a challenging task.

Researchers want to design re-ranking methods for image search to improve accuracy. According to previous research, there are mainly two types of re-ranking: global feature-based re-ranking and local feature-based re-ranking. Furthermore, both types of re-ranking methods could improve the accuracy of the image search engine. However, previous re-ranking methods are not fast or accurate enough. We implement novel global and local feature-based re-ranking methods to improve the accuracy of image search significantly. We test our re-ranking methods on popular image search databases. Experiments show that our re-ranking methods improve accuracy while slightly increasing computation time.

District heating systems (DHSs) have the potential to play a big part in the energy transition. The efficient operation of DHSs is therefore also an important subject of study. The operation of DHSs where combined heat and power (CHP) plants are used are particularly interesting, because CHPs can operate with high efficiency.

In this work, the operational optimization of DHSs with CHP plants is considered. Determining the optimal heat and electricity production for CHPs for multiple time steps into the future is a complex problem. Because of the heat storage capabilities in the network many solutions are feasible, but determining which solutions are infeasible because of constraint violations in the DHS involves computing time delays that depend on complex network dynamics.

In this work, the possibility of using an input convex neural network (ICNN) to learn the network dynamics of a DHS is explored. ICNNs have limitations on their learning capabilities, but theoretically allow for easier optimization. Experiments on the learning capabilities of ICNNs reveal that caution should be used when they are used to learn non-convex constraints, as the accuracy of the ICNN highly depends on how non-convex the function is. Experiments on the feasible space of supply temperatures to a small district heating network (DHN) reveal that although the network does not provide the same flexibility as heat storage tanks, still some flexibility in the operation can be found. This is due to the fact that water with a higher supply temperature is consumed by consumers at a slower pace and this increases the time delay between production and consumption. Supply temperatures that follow can then be lowered if the increased time delay causes this water to arrive when the heat demands are lower.

In the experiments it was found that this flexibility in operation translates to non-convex areas in the feasible space. When this space would be learned by an ICNN, this space would be made convex. How much of the flexibility would be removed by doing this is yet unknown and could be researched in future work. Other future work can be done on safely learning non-convex constraints with an ICNN.