Context: The study and analysis of (meta-)genomics have been providing scientists with valuable insights into the functioning and composition of microbial communities. Latest advancements in next-gen and high throughput sequencing technologies have resulted in significant growth in the data produced and made available for further research. These advancements can help scientists dive deeper into the analysis of uncultivated microbial populations that may have important roles in their environments. Gap: However, analysis of such data requires multiple preprocessing and computational steps to interpret the microbial and genetic composition of samples. For most researchers, configuring these tools, linking them with advanced binning and annotation tools, and maintaining the provenance of the processing continues to be extremely challenging. Moreover, the most common issue with current practices of metagenomics is the reproducibility of the research due to the complexity of setup and configurations. Aim: Our aim is to get a big-picture understanding of the common practices and approaches for metagenomic analyses and to find out which ones are more often used by researchers and why. Further, to compare some of the existing tools and look into possibilities of developing and/or using a reproducible pipeline and give some general recommendations for it. Methods: For this purpose, three main methods were used. First, a literature survey was performed on metagenomic analysis approaches, methodologies, and tools. Next, researchers and scientists with different educational backgrounds active in this field were interviewed. Lastly, the process of pipeline construction and bottlenecks were evaluated through hands-on experience. Findings: By conducting this research, several common pitfalls and shortcomings of metagenomic analysis practices were identified. Since the expertise of most researchers in this field is lacking a fundamental computer science and programming background, very few would attempt developing a pipeline from scratch. Therefore, if instead, they would opt for using “ready-made” General Purpose Pipelines (GPP), they would also face various difficulties in setting up and configuring them to their needs. Also, it has been observed that many of the existing metagenomic tools are not developed and maintained according to computer science code production standards. Therefore, even the more popular tools can suffer from detrimental bugs that can render them broken and consequently deprecated. However, with the emergence of the new “all-in-one” interface-based online platforms such as Kbase.us that enable simple point-and-click set-up and sharing of workflow, there is hope for entering a new era of reproducible metagenomic analysis.

Mobile learning is growing in popularity due to its nature and the access to mobile devices: students of different ages can use their mobile phones to access any information almost anywhere. The popularity of mobile devices and the current growth of social media allow to apply mobile social media learning in higher education. This systematic review gives an overview of the current usage of mobile social media learning, the use cases, barriers, advantages and difference between personal social media and higher education apps within a learning context. The main findings of this paper indicate that social media is a good tool for collaborative purposes, communication purposes, resource sharing and accessing. However, social media use cases are limited at the current state, as higher education apps provide more specific and learn-based functionalities. Furthermore, using social media in a formal way helps to make the education process more personal by having a direct connection with instructors. However, there are still barriers that institutions have to overcome, which include the social media integration models into higher education. This paper, therefore, also discusses the existing integration frameworks and proposes an extension of its current model.

Identifying novel malware and their behaviour enables security engineers to prevent and protect users with devices on the network from attackers. MalPaCA is an algorithm that helps to understand the behaviours of the network traffic by clustering uni-directional network connections which can be analyzed further to interpret which label suites the malicious connection. When clustering connections, features extracted from the packet information were chosen manually based on the generalizability of information and research of common malware characteristics. The feature set can be extracted automatically with an autoencoder to increase the representation of each packets in network traffics. A comparison with an autoencoder generated feature set to the hand-crafted feature set shows that the hand-crafted feature set represents the malicious traffics with higher accuracy and more insightful explainability. A comparative experiment is run on the IoT-23 dataset, a network traffic capture from Avast’s AIC laboratory.

Real-time sequence clustering is the problem of clustering an infinite stream of sequences in real time with limited memory. A variant of the k-medoids algorithm called SeqClu is the suggested approach, representing a cluster with p most representative sequences of the cluster, called prototypes, to solve the problem of maintaining a high-quality representation of a cluster that requires little memory throughout time. However, the computational cost of this algorithm is considerable due to many distance computations that use Dynamic Time Warping (DTW), which is a computationally expensive distance measure that can be applied to sequences and is proven to be robust to noise and
delays. Therefore, this paper proposes an extension of SeqClu called SeqClu-PV, characterised by a decision-making mechanism for updating prototypes that improves the balance between the number of distance computations and the cost incurred due to incorrect clustering and reviews its performance.

Malware Packet-sequence Clustering and Analysis (MalPaCA) is a unsupervised clustering application for malicious network behavior, it currently uses solely sequential features to characterize network behavior. In this paper an extensive comparison between those features and statistical features is performed. During the comparison a better clustering performance achievable with statistical features for longer connection sequences is shown and advice on which features can be added to MalPaCA.

Sustained attention is a cognitive state where the learners’ attention is completely focused on the learning environment and content-related thoughts for a continuous stretch of time. Sustained attention is vital to perform well on learning tasks, such as reading. Learning analytics platforms that detect changes in sustained attention can prevent ineffective learning by providing direct feedback to the learner. Prior research has found that eye gaze and blink patterns can be good indicators of cognitive state. In this paper we investigate the following main research question: "How can webcam-based eye gaze and blink pattern tracking indicate changes in learners' sustained attention in the remote learning context?". While other studies rely on expensive eye trackers to perform detection, this research explores the use of widely used laptop webcams for detecting changes in sustained attention. We collected webcam data through a small case study involving several different reading tasks. A machine learning classification model was trained on the collected webcam data. The resulting detection model performs well on validation data with a F1-score of 0.889. The model does not perform well on testing data however, showing that it is not usable in practice. We give several possible explanations for this behavior, most of them originating from an overfitted model due to the small size of the user study. Our findings indicate that future work should focus on different experimental settings and larger user studies.

The focus in over two decades of mobile learning (m-learning) research has meanly been on PK-12 education. M-learning is seen as an important new development in PK-12 and higher education (HE). Yet, the ease of app identification is lacking. This paper aimed to classify different m-learning applications to ensure that they are more easily accessible to teachers. Such a framework has been proposed earlier. However, the technological approach misses, and the apps are mainly for PK-12 pupils. Thus, the current paper aimed to create a framework for (learning) apps for higher education, including their pedagogical and technological approaches. Overall, eleven different app types were identified and categorized into four main categories: practice, organization, social, and creation. Then the apps were further classified using their pedagogical design and instructional approach. The applications can be identified as skill-, content- or function-based applications with overlap in the different categories. Finally, the mobile platform and the features used were analyzed and included in the framework here it was found that most of the apps were hybrid applications which allow for a diverse set of features for the apps. Web based applications are the least prominent and provide the least amount of features.

MalPaCA makes use of unsupervised machine learning to provide malware capability assessment by clustering the temporal behaviour of malware network packet traces. A comparative analysis was performed on various clustering algorithms to determine the best clustering algorithm in terms of network behaviour discovery. The clustering algorithms included in the analysis were HDBSCAN, OPTICS, Agglomerative Hierarchical Clustering and K-medoids. Metrics that capture cluster separation, cohesion, purity and completeness were used to determine the performance of the clustering algorithms. Agglomerative Hierarchical Clustering had the lowest total error of 0.950 in the comparative analysis compared to the baseline HDBScan with an error of 1.381.

Remote learning, learning from home using online available materials, is becoming increasingly more common. This paper focuses on reading activities during remote learning. An important part of learning is keeping sustained attention on the learning materials, as a shift from sustained attention to internal thought or mind-wandering oftentimes decreases the learning performance. Yet this shift is very common during daily activities. It is therefore important for a remote E-learning system to be aware of these attention shifts and allow it to intervene when necessary. Prior research indicates that facial emotion detection and participant's body temperature are generally missing as data modalities. Especially facial emotion recognition may contribute to this area of research, as facial emotions may indicate a learner's affective state during the learning process, and detection of this can be done using a simple consumer laptop web camera. Therefore this paper aims to investigate the feasibility of using a standard laptop web camera to detect facial emotions and a cheap Infrared Temperature Sensor (IR-sensor) to detect the loss of sustained attention. Two experiments were conducted to gather attentive and inattentive data. From this data, features were extracted which were used in multiple machine learning models. The created Machine Learning (ML) models worked well on synthetic test and validation data, but they performed poorly in practice. Our main hypothesis for this is overfitting of the ML models on the data, as due to the Coronavirus Disease (COVID-19) no more than three participants partook in the user studies. Overall, the conclusion is that both emotional facial detection and participant body temperature show great potential to detect sustained attention, but further research needs to be done with a larger group of participants to confirm this.

The mobile device market has been steadily growing throughout the last two decades. Nowadays, most people own smartphones that support various applications. Many of these applications fall under the category of mobile learning. These apps have many exciting features which could potentially enhance the learning experience in higher education. If academic institutions decide to use such applications, they also need instructions on how to integrate these apps in a common learning portal, such as a learning management system. This research identifies six features of modern mobile learning applications - augmented reality, gamification, artificial intelligence, push notifications, personalization, and collaborative learning. Then, it describes them and their applicability in both educational and non-educational contexts, with the focus on higher education. Finally, two technologies, namely learning tools interoperability (LTI) and application programming interfaces (APIs), are presented for data and application integration.

In this research, a learner’s sustained attention in the remote learning context will be studied by collecting data from different sensors. By combining the results of these sensors in a multi-modal analytics tool, the estimation of the learner’s sustained attention can hopefully be improved. This research will mainly focus on microphone recordings of ambient sound in a learners room. The main research question of this research was "How can ambient noise sensing aid in a multi-modal analytics tool to track sustained attention?". The multi-modal learning analytics tool, if accurate enough, could potentially be used by teachers to make their material more engaging and could help learner’s to keep their focus while performing a learning task (Schneider et al., 2015). The research resulted in a model with 61% accuracy. This percentage needs to be further researched, since because of the COVID situation, not enough data could be collected to train the model. Because of the relatively low accuracy of the model, it was found that ambient noise sensing can aid the multi-modal analytics tool to some extent by adding some data-points it is certain about, when the mobile movement tracking model does not detect a distraction. If the model improves in future research, the model could be able to help mobile movement tracking model, even if the mobile movement tracking model already predicts a distraction with bigger then 50% certainty.

Clustering is a group of (unsupervised) machine learning algorithms used to categorize data into clusters. The most popular clustering algorithm is k-means clustering. K-means clustering clusters the data into k clusters where a cluster is represented by the mean of the data points called a centroid. Instead of using the mean as a centroid, a data point (medoid) can be used instead. This algorithm is called k-medoids algorithm. Both the algorithms work in an offline setting where all the data is known beforehand and usually use Euclidean distance to calculate the distance between any two points. SeqClu is another clustering algorithm for sequential data that works in an online setting and uses Dynamic Time Warping as its distance measure. It is based on k-medoids where it uses p sequences called prototypes, to represent a cluster. It assigns an incoming sequence to the cluster that has the lowest average distance between its prototypes and the incoming sequence. The issue in this approach is that many Dynamic Time Warping distance calculations need to be made which affects the clustering speed and using the average distance affects the clustering accuracy due to outliers being assigned as prototypes. This paper proposes an alternative algorithm with three variants for the cluster assignment process. This algorithm iterates through the prototypes in search for the closest prototype while excluding clusters that are deemed too far. It assigns the incoming sequence to the cluster of the closest prototype that it has found. Experiments on the UJI Pen Characters and UCR Synthetic Control datasets show an improvement in clustering speed and accuracy.

One of the most important bottlenecks that contributes to the congestion of traffic is nonoptimal traffic signal control. Techniques that have been investigated to optimise traffic signal control have been focused on improving the traffic flow through individual intersections. However, if intersections are optimised based on only reducing local congestion, this could result in introducing congestion in other places. Therefore, this paper introduces a normal form game that can be used to analyse the impact on local and global performance when optimising traffic signal control. Results obtained from a realistic traffic simulation suggest that it is sometimes possible for each player to choose a traffic signal control strategy that optimises its own welfare but also maximises the social welfare. The results also indicate that sub-optimal traffic signal control strategy profiles can become optimal when exposed to certain traffic intensities.

Identifying the most efficient exploration approach for deep reinforcement learning in traffic light control is not a trivial task, and can be a critical step in the development of reinforcement learning solutions that can effectively reduce traffic congestion. It is common to use baseline dithering methods such as $\epsilon$-greedy. However, the value of more evolved exploration approaches in this setting has not yet been determined. This paper addresses this concern by comparing the performance of the popular deep Q-learning algorithm using one baseline and two state of the art exploration approaches, and their combination. Specifically, $\epsilon$-greedy is used as a baseline, and compared to the exploration approaches Bootstrapped DQN, randomized prior functions, and their combination. This is done in three different traffic scenarios, capturing different traffic profiles. The results obtained suggest that the higher the complexity of the traffic scenario, and the larger the size of the observation space of the agent, the larger the gain from efficient exploration. This is illustrated by the improved performance observed in the agents using efficient exploration and enjoying a larger observation space in the complex traffic scenarios.

Traffic congestion is a problem of tremendous size that affects many people. Using Reinforcement Learning to find a light control policy can ease traffic congestion and decrease travel time for vehicles. This paper specifically looks at the effect of using different reward functions for training agents. We highlight how the learnabilty of a reward function and its alignment with the final goal of the agent are the most important factors when designing a reward definition. Finally we propose a reward function to use for future

Cityhunter developed the idea of transforming their card game, Hunting Happy Families, into a game that can be played on tablets in a same manner that their other game, Hunt the Hunter, can be played. The aim of this project is to develop a game that can be played in Schiedam on an Android platform. An additional requirement is that the game can be adapted after the project is finished. That is, the client must be able to add new card sets and new locations through an interface without having to make changes in the codebase. Another part of this project includes a gamemaster interface, in which one person can keep track of all teams in a game, including their cards, points, locations and chat messages. Lastly, a server was created for storing and sending information between teams and the game master. This report defines the standard game play of a game of Hunting Happy Families. The main goal of the game is to learn about the history of the city of Schiedam. By walking around, teams can collect cards from other players, and identify locations associated with the various sets to try and gather as many points as possible. The team with the most points at the end of the game, wins. The system features a server connecting the teams and the game master using a string message system and lambda functions. Upon sending a message using this system, the server handles the requested data, and answers the request by forwarding the provided information, or relaying the information requested. The system was thoroughly field tested, including a playtest with external players. The verdict on the game by these players is positive, though the system needs a slight amount of fine tuning to ensure the best experience. That said, the game is completely ready for use, andHuntingHappy Families is bound to carve its niche in the market of group events.

The department of Urbanism at the TU Delft, our clients, research the sentiment in different places, times, ages or genders and compare them to each other. This report describes the purpose, design, implementation and accuracy of a web tool created to get insights into the sentiment people have, de- ducted from social media. The aim of our project was to make research easy and extracting innovative insights from social media.

In our tool, we analysed tweets and their location to collect information about the sentiment different people have towards places. The implementation of our tool consisted of five main components: (1) Twitter-Kafka, processing the tweets from the tweet data stream to our database, (2) face recognition, used for determining whether a tweet comes from a person instead of a company or organisation and for age and gender inference, (3) sentiment analysis, using machine learning to determine whether a tweet is neutral, negative or positive, (4) REST API, for the connection between the front-end and the back-end and (5) the user interface, in the form of an interactive dashboard.

At the beginning of the project, we set up a pipeline that checks the code on multiple things. The testing of the back-end is based on a Python unit test suit. For the build to succeed, all tests must pass and the total branch coverage must be at least 80%. We used Flake8 and ESlint in our build to ensure code quality at all times.

All of the above-mentioned components are up and running. The clients are now able to research the sentiments of people towards places.

Team Sunweb, a professional cycling team and our client, is constantly looking for innovations to help them win races. They tasked us with creating an application which could assist coaches with determining the strategy during a race. This application, which we dubbed TelaSol, is supposed to run on a tablet that is mounted on the dashboard inside the coach car. For this project we developed an application that allows races to be prepared on a desktop computer and tracked during a race on a tablet-optimized interactive dashboard. On this dashboard, there will be information on the riders, the route and comments that can be added before the race.

During development we have considered existing solutions, relevant literature and useful technologies to get an idea of what was possible and how we could achieve our goal. We used this knowledge to create our initial set of requirements. We then proceeded development of application using an agile approach, which involves regular feedback moments from our client to update the requirements and adjust our focus accordingly. To verify the quality of our product we relied on a combination of automated tests, user testing and validation through the client.

Initially the application was supposed to integrate live data coming from the riders during the race, but due to a regulation change we had to change our focus. Instead, we focused primarily on creating the application for playback purposes, while still keeping it adaptable to live data. The application performs the main tasks that were initially defined properly. After further development on live data and extensive situational testing, the app can be used to its full potential. Using TelaSol, Team Sunweb will improve their ability to analyze races and increase their chances of winning.