Change search
Refine search result
1234567 1 - 50 of 893
CiteExportLink to result list
Permanent link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Rows per page
  • 5
  • 10
  • 20
  • 50
  • 100
  • 250
Sort
  • Standard (Relevance)
  • Author A-Ö
  • Author Ö-A
  • Title A-Ö
  • Title Ö-A
  • Publication type A-Ö
  • Publication type Ö-A
  • Issued (Oldest first)
  • Issued (Newest first)
  • Created (Oldest first)
  • Created (Newest first)
  • Last updated (Oldest first)
  • Last updated (Newest first)
  • Disputation date (earliest first)
  • Disputation date (latest first)
  • Standard (Relevance)
  • Author A-Ö
  • Author Ö-A
  • Title A-Ö
  • Title Ö-A
  • Publication type A-Ö
  • Publication type Ö-A
  • Issued (Oldest first)
  • Issued (Newest first)
  • Created (Oldest first)
  • Created (Newest first)
  • Last updated (Oldest first)
  • Last updated (Newest first)
  • Disputation date (earliest first)
  • Disputation date (latest first)
Select
The maximal number of hits you can export is 250. When you want to export more records please use the Create feeds function.
  • 1.
    Aarflot, Ludvig
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Implementation of High Current Measurement Technology for Automotive Applications in Programmable Logic2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    At Inmotion Technologies, a special method of measuring phase currents is usedin the high power inverters for automotive applications. This method requiresa considerable amount of control logic, currently implemented with discretelogic gates distributed over a number of integrated circuits. In this thesis, thefeasibility of replacing this with programmable logic hardware in one singlepackage is investigated.The theory behind the current measurement method as well as the operationof the discrete implementation is analysed and described. Requirements ona programmable logic device to implement this was identified and a suitabledevice chosen accordingly. A prototype was developed and tested, interfacingan existing product.Benefits in terms of cost and size are evaluated as well as required changesto the existing system and the possibility for improvements brought by such achange is analysed. Since the products in question have high requirements onfunctional safety, possible impacts in this regard are discussed.

  • 2.
    Abdelmassih, Christian
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Container Orchestration in Security Demanding Environments at the Swedish Police Authority2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The adoption of containers and container orchestration in cloud computing is motivated by many aspects, from technical and organizational to economic gains. In this climate, even security demanding organizations are interested in such technologies but need reassurance that their requirements can be satisfied. The purpose of this thesis was to investigate how separation of applications could be achieved with Docker and Kubernetes such that it may satisfy the demands of the Swedish Police Authority.

    The investigation consisted of a literature study of research papers and official documentation as well as a technical study of iterative creation of Kubernetes clusters with various changes. A model was defined to represent the requirements for the ideal separation. In addition, a system was introduced to classify the separation requirements of the applications.

    The result of this thesis consists of three architectural proposals for achieving segmentation of Kubernetes cluster networking, two proposed systems to realize the segmentation, and one strategy for providing host-based separation between containers. Each proposal was evaluated and discussed with regard to suitability and risks for the Authority and parties with similar demands. The thesis concludes that a versatile application isolation can be achieved in Docker and Kubernetes. Therefore, the technologies can provide a sufficient degree of separation to be used in security demanding environments.

  • 3.
    Abgrall, Corentin
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Deep learning models as advisors to execute trades on financial markets2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Recent work has shown that convolutional networks can successfully handle time series as input in various different problems. This thesis embraces this observation and introduces a new method combining machine learning techniques in order to create profitable trading strategies. The method addresses a binary classification problem: given a specific time, access to prices before this moment and an exit policy, the goal is to forecast the next price movement. The classification method is based on convolutional networks combining two major improvements: a special form of bagging and a weight propagation, to enhance the accuracy and reduce the overall variance of the model. The rolling learning and the convolutional layers are able to exploit the time dependency to strongly improve the trading strategy. The presented architecture is able to surpass the expert traders.

  • 4.
    Aboode, Adam
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Anomaly Detection in Time Series Data Based on Holt-Winters Method2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    In today's world the amount of collected data increases every day, this is a trend which is likely to continue. At the same time the potential value of the data does also increase due to the constant development and improvement of hardware and software. However, in order to gain insights, make decisions or train accurate machine learning models we want to ensure that the data we collect is of good quality. There are many definitions of data quality, in this thesis we focus on the accuracy aspect.

    One method which can be used to ensure accurate data is to monitor for and alert on anomalies. In this thesis we therefore suggest a method which, based on historic values, is able to detect anomalies in time series as new values arrive. The method consists of two parts, forecasting the next value in the time series using Holt-Winters method and comparing the residual to an estimated Gaussian distribution.

    The suggested method is evaluated in two steps. First, we evaluate the forecast accuracy for Holt-Winters method using different input sizes. In the second step we evaluate the performance of the anomaly detector when using different methods to estimate the variance of the distribution of the residuals. The results indicate that the suggested method works well most of the time for detection of point anomalies in seasonal and trending time series data. The thesis also discusses some potential next steps which are likely to further improve the performance of this method.

  • 5.
    Abrahamsson, Felix
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Designing a Question Answering System in the Domain of Swedish Technical Consulting Using Deep Learning2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Question Answering systems are greatly sought after in many areas of industry. Unfortunately, as most research in Natural Language Processing is conducted in English, the applicability of such systems to other languages is limited. Moreover, these systems often struggle in dealing with long text sequences.

    This thesis explores the possibility of applying existing models to the Swedish language, in a domain where the syntax and semantics differ greatly from typical Swedish texts. Additionally, the text length may vary arbitrarily. To solve these problems, transfer learning techniques and state-of-the-art Question Answering models are investigated. Furthermore, a novel, divide-and-conquer based technique for processing long texts is developed.

    Results show that the transfer learning is partly unsuccessful, but the system is capable of perform reasonably well in the new domain regardless. Furthermore, the system shows great performance improvement on longer text sequences with the use of the new technique.

  • 6.
    Ackland, Marcus
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Investigating Adaptive Trajectories to Explore Water Plumes on Icy Moons2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    This thesis investigates different means of introducing autonomy intodesigning spacecraft trajectories by surveying four different adaptivetrajectories. This is done using established algorithms within pathplanning and robotics, implementing trajectories based on splines andgreedy algorithms. The survey is based on real planetary data lettingthe spacecraft fly through the water plumes on Enceladus. Theseplumes are constructed using analytical models of the water plumesthat have been fitted to measurements made by Cassini during flybysover the last decade. A mission designed to probe these plumes at analtitude as low as 5 km is studied as a test scenario for the adaptive trajectories.It is found that the adaptive trajectories increase the sciencereturn at lower altitudes, both in exploring the terrain and samplingthe plume material.

  • 7.
    Ackva, Adrian
    KTH, School of Electrical Engineering and Computer Science (EECS).
    WinBro: A Window and Broadcast-based Parallel Streaming Graph Partitioning Framework for Apache Flink2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The past years have shown an increasing demand to process data of various kinds and size in real-time. A common representation for many real-world scenarios is a graph, which shows relations between entities, such as users of social networks or pages on the Internet. These graphs increase in size over time and can easily exceed the capacity of single machines.Graph partitioning is used to divide graphs into multiple subgraphs on different servers. Traditional partitioning techniques work in an offline manner where the whole graph is processed before partitioning. Due to the recently increased demand for real-time analysis, online partitioning algorithms have been introduced. They are able to partition a graph arriving as a stream, also referred to as a streaming graph, without any pre-processing step.The goal of a good graph partitioning algorithm is to maintain the data locality and to balance partitions’ load at the same time. Although different algorithms have proven to achieve both goals for real-world graphs, they often require to maintain a state. However, modern stream processing systems, such as Apache Flink, work with a shared-nothing architecture in a data-parallel manner. Therefore, they do not allow to exchange information along with parallel computations. These systems usually use Hash-based partitioning, that is a fast stateless technique but ignores the graph structure. Hence, it can lead to longer analysis times for streaming applications which could benefit from preserved structures.This work aims to develop a state-sharing parallel streaming graph partitioner for Apache Flink, called WinBro, implementing well-performing partitioning algorithms. In order to do this, existing streaming graph algorithms are studied for possible implementation and then integrated into WinBro.For validation, different experiments were made with real-world graphs. In these experiments, the partitioning quality, and partitioning speed were measured. Moreover, the performance of different streaming applications using WinBro was measured and compared with the default Hash-based partitioning method.Results show that the new partitioner WinBro provides better partitioning quality in terms of data locality and also higher performance for applications with requirements for locality-based input data. Nonetheless, the Hash-based partitioner shows the highest throughput and better performance for data localityagnostic streaming applications.

  • 8.
    Adam, Jonathan
    KTH, School of Electrical Engineering and Computer Science (EECS), Media Technology and Interaction Design, MID.
    Analyzing Function and Potential in Cuba's El Paquete: A Postcolonial Approach2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The dire state of Cuban internet connectivity has inspired local informal innovations. One such innovation is El Paquete, a weekly distribution of downloaded content spread through an informal network. Taking a postcolonial approach, I investigate through user experiences how this network operates in a resource-poor environment. This investigation articulates a model of El Paquete centered on social interactions, which inform the system’s function but also shape El Paquete’s design and role in society. Based on this model, a set of speculative design exercises probe possibilities to streamline El Paquete’s compilation, involve consumer preferences in its design directions, or act as a disruption tolerant network. In uncovering the technical possibilities of El Paquete, these designs illuminate how its current design serves Cuban communities by embodying realities and limitations of Cuban society. El Paquete’s embodiment of informal innovation serves as a call to designers to continuously rethink development design processes, centering communities and their knowledge and technical practices.

  • 9.
    Adamsson, Marcus
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Curriculum learning for increasing the performance of a reinforcement learning agent in a static first-person shooter game2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    In this thesis, we trained a reinforcement learning agent using one of the most recent policy gradient methods, proximal policy optimization, in a first-person shooter game with a static player. We investigated how curriculum learning can be used to increase performance of a reinforcement learning agent. Two reinforcement learning agents were trained in two different environments. The first environment was constructed without curriculum learning and the second environment was with curriculum learning. After training the agents, the agents were placed in the same environment where we compared them based on their performance. The performance was measured by the achieved cumulative reward. The result showed that there is a difference in performance between the agents. It was concluded that curriculum learning can be used to increase the performance of a reinforcement learning agent in a first-person shooter game with a static player.

  • 10.
    Adaszynski, Wojciech
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Interactive visualization of radio waves propagation in 5G massive MIMO2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The complexity of advanced antenna techniques used in the new generation of wireless networks (5G) makes communication between experts and non-technical staff more difficult than ever. As cooperation between network vendors and network operators affects the adoption of the new standard, a need for a new tool has emerged to make technical presentations more engaging and compelling. This thesis presents an exploratory study that aims to examine various design options for an interactive visualization of radiowave propagation to be used by advanced antenna systems experts. Through a Research-oriented Design, functional and non-functional requirements were identified with the help of domain expert. Later, an interactive prototype was designed and developed using a participatory design approach. Qualitative and quantitative data was gathered through usability testing, System Usability Scale (SUS) questionnaires and semi-structured interviews conducted with 12 researchers and engineers at Ericsson AB a multinational telecommunication company. User evaluation proved that such a tool could facilitate communication between technical experts and non-technical staff. The developed prototype was considered intuitive and useful by the majority of study participants as measured by interviews and the SUS survey. Future research is encouraged to include the target audience representatives in order to measure their engagement while using the tool.

  • 11.
    Aditya Wardana, I Wayan Kurniawan
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Visualizing Error in Real-Time Video Streaming Data for a Monitoring System2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The aim of this master thesis is to investigate the affordances and limitations of using information visualization methods to visualize errors in real-time video streaming data. The study was carried in Red Bee Media company by following several steps including user research, prototyping, and user evaluation. The user research produced design requirements and basic tasks for the prototype. The prototype had to follow the design requirements and use information visualization techniques to visualize the error data. Next, the prototype was evaluated by 5 expert users, all Red Bee Media employees with 1,5 to 3 years experience of working with the existing Red Bee Media system. The results show the prototype obtained a higher SUS score compared to the Red Bee Media monitoring system. Based on a comparison questionnaire, the prototype also had a better visualization for each basic task compared to Red Bee Media monitoring system. The comments from the user evaluation have been categorized into 4 different labels. Those labels listed several usabilities need to be focused on when developing a video monitoring system.

  • 12.
    Adolfsson, Fredrik
    KTH, School of Electrical Engineering and Computer Science (EECS).
    WebTaint: Dynamic Taint Tracking for Java-based Web Applications2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The internet is a source of information and it connects the world through a single platform. Many businesses have taken advantage of this to share information, to communicate with customers, and to create new business opportunities. However, this does not come without drawbacks as there exists an elevated risk to become targeted in attacks.

    The thesis implemented a dynamic taint tracker, named WebTaint, to detect and prevent confidentiality and integrity vulnerabilities in Java-based web applications. We evaluated to what extent WebTaint can combat integrity vulnerabilities. The possible advantages and disadvantages of using the application is introduced as well as an explication whether the application was capable of being integrated into production services.

    The results show that WebTaint helps to combat SQL Injection and Cross-Site Scripting attacks. However, there are drawbacks in the form of additional time and memory overhead. The implemented solution is therefore not suitable for time or memory sensitive domains. WebTaint could be recommended for use in test environments where security experts utilize the taint tracker to find TaintExceptions through manual and automatic attacks.

  • 13.
    Adrup, Joakim
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Visualization and Interaction with Temporal Data using Data Cubes in the Global Earth Observation System of Systems2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The purpose of this study was to explore the usage of data cubes in the context of the Global Earth Observation System of Systems (GEOSS). This study investigated what added benefit could be provided to users of the GEOSS platform by utilizing the capabilities of data cubes. Data cubes in earth observation is a concept for how data should be handled and provided by a data server. It includes aspects such as flexible extraction of subsets and processing capabilities. In this study it was found that the most frequent use case for data cubes was time analysis. One of the main services provided by the GEOSS portal was the discovery and inspection of datasets. In the study a timeline interface was constructed to facilitate the exploration and inspection of datasets with a temporal dimension. The datasets were provided by a data cube, and made use of the data cubes capabilities in retrieving subsets of data along any arbitrary axis. A usability evaluation was conducted on the timeline interface to gain insight into the users requirements and user satisfaction. The results showed that the design worked well in many regards, ranking high in user satisfaction. On a number of points the study highlighted areas of improvement. Providing insight into important design limitations and challenges together with suggestions on how these could be approached in different ways.

  • 14.
    Afework, Miriam
    KTH, School of Electrical Engineering and Computer Science (EECS), Media Technology and Interaction Design, MID.
    Using Magic Machines to Elaborate Menstrual Self-Tracker­s for Women with Endometriosis2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Existing self-tracking tools for women concentrate on one’s general well-being and keeping track of ovulation and periods. With around 10% of women worldwide suffering from endometriosis there is an unmet need to leverage self-tracking for women whose cycles are affected by more variables. The disease is enigmatic with an unknown cause and cure and the ill­ness differs for each individual in symptoms and working treatments. It is therefore critical to understand how women can learn about their bodies and how to treat their condition. In this research I work with two sufferers to identify their secret de­sires through a workshop and a series of interviews. Results suggest that women with endometriosis could benefit from ex­perimenting with different habits and make personalized routines to suit their own needs. Finally I present design implica­tions for an existing menstrual app in the form of an add-on. The steps of the add-on tool included three steps. Firstly, choosing variables of one’s well being to track (mood, energy, pain etc.), choosing activities for one or more cycles (gluten-free diet, exercising etc.), and lastly viewing an analysis of any changes in the body.

  • 15.
    Ahlström, Marcus
    KTH, School of Electrical Engineering and Computer Science (EECS), Media Technology and Interaction Design, MID.
    Broadening the Reading Experience on Mobile Devices using Tilt-based Input: An Explorative Design Study2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    This thesis is an explorative study aimed at the possibility of integrating tilt-based input to improve the reading experience on smartphones. Previous works from the early 2000s have been skeptical towards tilt-based navigation, deeming it unruly and imprecise. To investigate if today’s technology has unlocked new possibilities; two experimental reading methods were designed, created and tested iteratively on 20, respectively 18 participants. The first method is a reassessment of tilt-based auto-scrolling and the second is a novel approach comparable to tilt-based paging. Data from the reading sessions were collected quantitatively in tandem with qualitative data from post-session interviews. The results indicate good potential and a reading performance similar to the standard navigation method. The importance of accommodating people with different reading behaviours was also discussed.

  • 16.
    Aksjonova, Jevgenija
    KTH, School of Electrical Engineering and Computer Science (EECS).
    LDD: Learned Detector and Descriptor of Points for Visual Odometry2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Simultaneous localization and mapping is an important problem in robotics that can be solved using visual odometry -- the process of estimating ego-motion from subsequent camera images. In turn, visual odometry systems rely on point matching between different frames. This work presents a novel method for matching key-points by applying neural networks to point detection and description. Traditionally, point detectors are used in order to select good key-points (like corners) and then these key-points are matched using features extracted with descriptors. However, in this work a descriptor is trained to match points densely and then a detector is trained to predict, which points are more likely to be matched with the descriptor. This information is further used for selection of good key-points. The results of this project show that this approach can lead to more accurate results compared to model-based methods.

  • 17.
    Al Hakim, Ezeddin
    KTH, School of Electrical Engineering and Computer Science (EECS).
    3D YOLO: End-to-End 3D Object Detection Using Point Clouds2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    For safe and reliable driving, it is essential that an autonomous vehicle can accurately perceive the surrounding environment. Modern sensor technologies used for perception, such as LiDAR and RADAR, deliver a large set of 3D measurement points known as a point cloud. There is a huge need to interpret the point cloud data to detect other road users, such as vehicles and pedestrians.

    Many research studies have proposed image-based models for 2D object detection. This thesis takes it a step further and aims to develop a LiDAR-based 3D object detection model that operates in real-time, with emphasis on autonomous driving scenarios. We propose 3D YOLO, an extension of YOLO (You Only Look Once), which is one of the fastest state-of-the-art 2D object detectors for images. The proposed model takes point cloud data as input and outputs 3D bounding boxes with class scores in real-time. Most of the existing 3D object detectors use hand-crafted features, while our model follows the end-to-end learning fashion, which removes manual feature engineering.

    3D YOLO pipeline consists of two networks: (a) Feature Learning Network, an artificial neural network that transforms the input point cloud to a new feature space; (b) 3DNet, a novel convolutional neural network architecture based on YOLO that learns the shape description of the objects.

    Our experiments on the KITTI dataset shows that the 3D YOLO has high accuracy and outperforms the state-of-the-art LiDAR-based models in efficiency. This makes it a suitable candidate for deployment in autonomous vehicles.

  • 18.
    Alam, Samiul
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Recurrent neural networks in electricity load forecasting2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    In this thesis two main studies are conducted to compare the predictive capabilities of feed-forward neural networks (FFNN) and long short-term memory networks (LSTM) in electricity load forecasting.

    The first study compares univariate networks using past electricity load, as well as multivariate networks using past electricity load and air temperature, in day-ahead load forecasting using varying lookback periods and sparsity of past observations. The second study compares FFNNs and LSTMs of different complexities (i.e. network sizes) when restrictions imposed by limitations of the real world are taken into consideration.

    No significant differences are found between the predictive performances of the two neural network approaches. However, adding air temperature as extra input to the LSTM is found to significantly decrease its performance. Furthermore, the predictive performance of the FFNN is found to significantly decrease as the network complexity grows, while the predictive performance of the LSTM is found to increase as the network complexity grows. All the findings considered, we do not find that there is enough evidence in favour of the LSTM in electricity load forecasting.

  • 19.
    Albrecht, Tomás
    KTH, School of Electrical Engineering and Computer Science (EECS), Media Technology and Interaction Design, MID.
    Designing the Publikvitto, a system to make government expenditure tangible2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Air transportation is essential to our society. It enables global trading, brings people together, and lets travelers explore distant parts of the world. However, flying is a highly unsustainable behavior and accounts for roughly 2% of all carbon emissions; with industry and research forecasting constant growth in the coming years. The economic benefits rhetoric often prevails over the environmental costs, though; motivating governments to give incentives to airports and airlines. The Swedish Government, despite its green goals and pro-sustainability actions, is no exception, and both municipal and federal funds support the air route network.

    This thesis reports on the development of the Publikvitto, a system designed to help citizen make sense of the government's incentives to the flying industry. The process is based on research through design and inspired by reflective practices. The primary outcome are insights into the relationship between designer, social issues, and government's actions; and how these elements can be approached in order to design artifacts that motivate people to engage in political discussions.

  • 20.
    Aleksandrauskaite, Ruth
    KTH, School of Electrical Engineering and Computer Science (EECS), Automatic Control.
    Analysis of Velocity Estimation Methods for High-Performance Motion Control Systems2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The majority of all commercial electronics hardware is manufactured usingSurface Mount Technology (SMT). Nevertheless, the increased complexityand miniaturization of electronics impose tough performance requirementson the automation process.The research in this paper concerns test and analysis of alternative velocityestimation methods for high-performance embedded motion control systems.The motion system in Mycronic’s pick and place machines is regulated by amotion controller consisting of a feedforward component and a feedback controller.The linear displacement is measured with an incremental encoder andthe velocity is estimated with a state observer. Previous work suggests thatthe velocity estimation is inadequate.Different observer designs including state and disturbance estimators weretested and evaluated through simulations in MATLAB SIMULINKr. Afterthat, experiments were performed on a conveyor retrieved from a pick andplace machine.The results show that a Kalman filter is the best state estimator. However,the method requires extensive tuning to attain good performance. The trackingperformance and robustness of the motion control system was highly improvedwhen using a Perturbation observer with Kalman filtering. Nonetheless,the settling time for point-to-point movements was somewhat shorterwhen using a Kalman filter alone.

  • 21.
    Alkeaid, Majed Mohammed G
    KTH, School of Electrical Engineering and Computer Science (EECS), Electric Power and Energy Systems. KTH, School of Industrial Engineering and Management (ITM).
    Study of NEOM city renewable energy mix and balance problem2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    It is important for NEOM management in the contemporary world to put in place NEOM projects using the available resources. The region in which the NEOM project is spacious and vast with conditions suited to generate energy from solar and wind. The NEOM projectis expected to be set up in the very resourceful state of Saudi Arabia. The purpose of the study is to assist in setting up a sustainable city through the exploitation of solar and wind energy. The aim of the study was to assist in the generation of more than 10 GW renewable energy to replace approximately 80,000 barrels of fossil energy. The problem of coming up with renewable and sustainable energy from the unexploited sources is addressed. The renewable city is expected to be a technological hub based on Green Energy with 100% renewable energy, which is correspond to 72:4GW. Freiburg and Masdar as renewable cities are used as case studies in the research. NEOM power generation capacity is capable to cover Saudi Arabia power generation capacity (approximately 71GW), which is more than enough for a city. The study reveals that the total power generation from wind farms, tidal farms, solar stations, and solar power tower stations are 9:1373GW, 4:76GW, 57:398GW and 1:11GW respectively. Saudi Arabia has plans to set up 16 nuclear plants (17 GW each) for energy purposes (total of 272 GW), which will be part of Saudi Arabia national grid and will be more than enough to cover NEOM electricity demand in case NEOM does not reach demand capacity. In case NEOM energy does not meet the demand, electricity generation from 16 Nuclearpower plants generating 17GW each, and 6 Natural underground batteries with a capacity of 120MW each are recommended. The study results can be applied in NEOM Institute of Science and Technology for further research on renewable energy. The findings can also be used for research extension of HVDC transmission lines between NEOM and Saudi Arabia main grid, Egypt, and Jordan.

  • 22.
    Alm, Erik
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Area and Power Efficiency of Multiplier-Free Finite Impulse Response Filters2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    In digital radio systems, a large number of finite impulse response filters are typically used. Due to their nature of operation, such filters require many multiplication operations, leading to great costs in terms of both chip area and power consumption. For cost reduction reasons, there is a strong business case for implementing these filters without general multipliers so as to reduce the area and power consumption of the overall system.This thesis explores a method of implementing finite impulse response halfband filters without general multipliers, by using a special filter structure and replacing multipliers with sequences of binary shifts and additions. The savings in terms of area and power consumption are estimated and compared to a conventional filter (with a common structure) implementation containing general multipliers, as well as the same conventional filter implemented without general multipliers by means of manipulating its coefficients such that they can be implemented with shifts and additions.The results show that while using the special filter structure with shifts and additions consumes less area and power than a conventional filter with general multipliers, employing simpler methods to obtain coefficients implementable with shifts and additions in a conventional filter structure produces smaller filters consuming less power. Moreover, the results of this thesis show that using methods allowing for multiplier-free filter implementations with conventional filter structures seems favorable, hence further investigation of such methods is recommended. Future studies could also focus on methods applicable to filters with support for dynamic coefficients.

  • 23.
    Almqvist, Andreas
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Sites of Encounter: Engagement Potentials and Considerations for Encompassing Respect2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    In this work, I address challenges of situated alienation from people and place. Using interaction design for placemaking, light is shed on a design space of social places with opportunities for planned and spontaneous activities to be done alone, with known people or with strangers. In research through design, four conceptual design instances were created, drawing from

    first-person and participatory perspectives, to unpack potentials for engagement and considerations for encompassing respect (meaning vigilant of neoliberal capitalist and masculine power relations). I contribute with sensitizing questions making interaction design aspects more accessible for designers entering this public design space.

  • 24.
    Alnesjö, Robert
    KTH, School of Computer Science and Communication (CSC).
    Chunked DASH in JavaScript2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Chunked DASH is getting attention for reducing otherwise high delay oflive segment streaming but there are a lot of unexplored problems associatedwith it. This master’s thesis investigates the difficulties involved withimplementing a chunked DASH player in the browser with JavaScript.

    A small system containing one JavaScript client and a server which simulateslive streaming by repeating VOD segments is implemented. Issuesrelated to the downloading of chunked segments are addressed and solvedsuch that chunked segments can be streamed within expected delay, andwith accurate throughput metrics.

  • 25.
    Al-qaysi, Ibrahim
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Communicating with a Smart Pillbox via Near Field Communication (NFC): A Mobile Application for Healthcare Professionals2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The lack of medication adherence leads to an incremental risk of diseases which can be a major burden on the individual, healthcare system, and society. Hence, healthcare professionals have a central role and should manage, guide, educate, and make their patient more involved in their treatment and thereby promoting a better medication adherence.

    Medication adherence is a great challenge for many patients with chronic conditions, elderly patient, or patient prescribed to long-term medication. The rapid development and deployment of mobile phones in the healthcare industry has an important role to play in this area and has led to the development of new phone features and applications that can help both caregivers and patients with managing and monitoring medication intakes. This development and support of mobile phones and applications have created and improved doctor-patient interaction.

    Today, there is no easy way for healthcare professionals to monitor and help patients with their medication intakes. A solution to this problem is to develop a mobile application that communicates with a smart pillbox via near field communication (NFC) to monitor, manage, and improve patient’s medication intakes in an easy and accessible manner. Using NFC as a communication technology allows data to be wirelessly transferred from phone to pillbox and vice versa. This solution will help healthcare professionals to create better treatment conditions and fewer side effects for their patients. These patients will be more knowledgeable and motivated to take greater responsibility in following doctor’s instructions, thereby improving their treatment process.

    The application is tested and evaluated during every iteration phase of the development process. These tests have been conducted by allowing healthcare professionals to test the application and provide feedback on their experience when using the app. Conducting these tests have helped with generating new ideas, features, and functionalities, but also helped to improve the user interface to make the application as user-friendly as possible.

  • 26.
    Alsing, Oscar
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Mobile Object Detection using TensorFlow Lite and Transfer Learning2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    With the advancement in deep learning in the past few years, we are able to create complex machine learning models for detecting objects in images, regardless of the characteristics of the objects to be detected. This development has enabled engineers to replace existing heuristics-based systems in favour of machine learning models with superior performance. In this report, we evaluate the viability of using deep learning models for object detection in real-time video feeds on mobile devices in terms of object detection performance and inference delay as either an end-to-end system or feature extractor for existing algorithms. Our results show a significant increase in object detection performance in comparison to existing algorithms with the use of transfer learning on neural networks adapted for mobile use.

  • 27.
    Alsterman, Marcus
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Transfer Learning for Sales Volume Forecasting Using Convolutional Neural Networks2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Improved time series forecasting accuracy can enhance demand planning, and therefore save money and reduce environmental impact. The idea behind this degree project is to explore transfer learning for time series forecasting. This has boiled down to two concrete goals. The first one is to examine if transfer learning can improve the forecasting accuracy when using a convolutional neural network (CNN) with dilated causal convolutions. The second goal is to investigate whether transfer learning makes it possible to forecast time series with less historical data.In this project, time series describing sales volume and price from three different consumer appliances are used. The length of the time series is about three years. Two transfer learning techniques are used: shared-hidden-layer CNN and pre-training. To tackle the first goal, the two transfer learning techniques are benchmarked against a CNN. The second goal is investigated conducting an experiment where the training set size varies for both a CNN and the two transfer learning techniques.The results from the first experiment indicate that transfer learning neither increase nor decrease forecasting accuracy. Interestingly, the second experiment however show that only 60 % (40 % for the SHL-CNN) of training samples is optimal for all models. This goes against the intuition that more training data leads to better model performance and this is most likely a phenomenum related specifically to time series forecasting. However, the percentage of 60 % most likely is application specific, we also find that pre-training, from any of the other products, improves the forecasting accuracy. Finally, reducing the training set further (20 % of training samples) affect the model differently. One pre-training model performs better than the rest, which perform very similar. This indicates that there are cases when transfer learning allows for forecasting smaller time series. However, further studies are required to establish how general these observations are.

  • 28.
    Al-Tai, Elias
    KTH, School of Electrical Engineering and Computer Science (EECS).
    An evaluation of the expressive power and performance of JSON-to-JSON transformation languages2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    JSON-to-JSON transformation languages enable the transformation of a JSON document into another JSON document. As JSON is gradually becoming the most used interchange format on the Internet there is a need for transformation languages that can transform the data stored in JSON in order for the data to be used with other systems. The transformation can transform the document structurally, for example by altering the hierarchical structure of the document. The transformation can also transform the document textually, for example by renaming fields or altering values. None of the existing JSON-to-JSON transformation languages have become a standard (Jellife, 2017). This work evaluates the expressive power of the JSON-to-JSON transformation language Jolt. Jolt have recently been adopted by Apache and support have been introduced in some of their products. If a transformation language have expressive power that are at least equal to Nested Relational Algebra this implies that a transformation language can perform many advanced transformations. In this work  a formal model of Jolt is defined, referred to as Jolt0, in order to compare its expressive powers to Nested Relational Algebra. For that purpose, the operations of another formal model called MQuery which have been proven to have equivalent expressive power to Nested Relational Algebra are translated into Jolt0. It is shown that Jolt does not have expressive powers equivalent to Nested Relational Algebra.

    We further compared the performance of four JSON-to-JSON transformation languages (Jolt, Handlebars, Liquid, and XSLT 3.0) by constructing tests where the different transformation languages executed equivalent transformations. The transformations were evaluated by measuring runtime and memory usage. The study shows that XSLT 3.0 performed worst in all run time and memory usage tests. When transforming large input data XSLT 3.0 performed significantly worse than the other languages.

  • 29.
    Alvarez Custodio, Maria
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Autonomous Recharging System for Drones: Detection and Landing on the Charging Platform2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    In the last years, the use of indoor drones has increased significantly in many different areas. However, one of the main limitations of the potential of these drones is the battery life. This is due to the fact that the battery size has to be limited since the drones have a maximum payload in order to be able to take-off and maintain the flight. Therefore, a recharging process need to be performed frequently, involving human intervention and thus limiting the drones applications.

    In order to solve this problem, this master thesis presents an autonomous recharging system for a nano drone, the Crazyflie 2.0 by Bitcraze AB. By automating the battery recharging process no human intervention will be needed, and thus overall mission time of the drone can be considerably increased, broadening the possible applications.

    The main goal of this thesis is the design and implementation of a control system for the indoor nano drone, in order to control it towards a landing platform and accurately land on it. The design and implementation of an actual recharging system is carried out too, so that in the end a complete full autonomous system exists.

    Before this controller and system are designed and presented, a research study is first carried out to obtain a background and analyze existing solutions for the autonomous landing problem.

    A camera is integrated together with the Crazyflie 2.0 to detect the landing station and control the drone with respect to this station position. A visual system is designed and implemented for detecting the landing station. For this purpose, a marker from the ArUco library is used to identify the station and estimate the distance to the marker and the camera orientation with respect to it.

    Finally, some tests are carried out to evaluate the system. The flight time obtained is 4.6 minutes and the landing performance (the rate of correct landings) is 80%.

  • 30.
    Amine Ramdani, Ahmed
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Rudnik, Sebastian
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Design and Construction of High Current Winding for a Transverse Flux Linear Generator Intended for Wave Power Generation2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    There is currently a high demand for electric power from renewablesources. One source that remains relatively untapped is the motionof ocean waves. Anders Hagnestål has been developing a uniquelyefficient and simplified design for a point-absorb buoy generator byconverting its linear motion directly into alternating electric power usinga linear PM engine. To test this method, a smaller prototype isbuilt. Its characteristics present some unusual challenges in the designand construction of its winding.Devices of this type typically use relatively low voltage (690V typicallyfor a wind turbine, compared to the 10kV range of traditionalpower plants). To achieve high power, they need high current, whichin turn requires splitting the conductors in the winding into isolatedparallel strands to avoid losses due to eddy currents and current crowding.However, new losses from circulating currents can then arise. Inorder to reduce said losses, the parallel conductors should be transposedin such a way that the aggregate electromotive force the circuitsthat each pair of them forms is minimized.This research and prototyping was performed in absence of advancedindustrial means of construction, with limited space, budget,materials, manpower, know-how, and technology. Manual ingenuityand empirical experimentation were required to find a practical implementationfor: laying the cables, fixing them in place, transferringthem to the machine, stripping their coating at the ends and establishinga reliable connection to the current source.Using theoretical derivations and FEM simulation, a sufficientlygood transposition scheme is proposed for the specific machine thatthe winding is built for. A bobbin replicating the shape of the enginecore is built to lay down the strands.The parallel strands are then organized each into their respectivebobbin, with a bobbin rack and conductor funneling device being designedand constructed to gather them together into a strictly-organizedbundle. An adhesive is found to set the cables in place.Problems with maintaining the orientation and configuration of thecables in the face of repeated torsion are met and solved. A chemicalsolution is used to strip the ends of the conductors, and a reliableconnection is established by crimping the conductors into a bi-metalCu-Al lug.ivIn conclusion, the ideal transposition schemes required to cancelout circulating currents due to magnetic flux leakage are impossibleto put in practice without appropriate technological means. The feasibletransposition scheme turns out to be a simple mirroring of conductors’positions, implemented by building each half of the windingseparately around replicas of the core and then connecting them usingcrimping lugs.

  • 31.
    Andersson, Morgan
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Personal news video recommendations based on implicit feedback: An evaluation of different recommender systems with sparse data2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The amount of video content online will nearly triple in quantity by 2021 compared to 2016. The implementation of sophisticated filters is of paramount importance to manage this information flow. The research question of this thesis asks to what extent it is possible to generate personal recommendations, based on the data that news videos implies. The objective is to evaluate how different recommender systems compare to complete random, each other and how they are received by users in a test environment.

    This study was performed during the spring of 2018, and explore four different algorithms. These recommender systems include a content-based, a collaborative-filter, a hybrid model and a popularity model as a baseline. The dataset originates from a news media startup called Newstag, who provide video news on a global scale. The data is sparse and includes implicit feedback only.

    Three offline experiments and a user test were performed. The metric that guided the algorithms offline performance was their recall at 5 and 10, due to the fact that the top list of recommended items are of most interest. A comparison was done on different amounts of meta-data included during training. Another test explored respective algorithms performance as the density of the data increased. In the user test, a mean opinion score was calculated based on the quality of recommendations that each of the algorithms generated for the test subjects. The user test also included randomly sampled news videos to compare with as a baseline.

    The results indicate that for this specific setting and data set, the content-based recommender system performed best in both the recall at five and ten, as well as in the user test. All of the algorithms outperformed the random baseline.

  • 32.
    Andre do Nascimento, Allan
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Robust Model Predictive Control for Marine Vessels2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    This master thesis studies the implementation of a Robust MPC controllerin marine vessels on different tasks. A tube based MPC is designed based onsystem linearization around the target point guaranteeing local input to statestability of the respective linearized version of the original nonlinear system.The method is then applied to three different tasks: Dynamic positioningon which recursive feasibility of the nominal MPC is also guaranteed, Speed-Heading control and trajectory tracking with the Line of sight algorithm.Numerical simulation is then provided to show technique’s effectiveness.

  • 33.
    Andruetto, Claudia
    KTH, School of Electrical Engineering and Computer Science (EECS).
    FEASIBILITY ANALYSIS OF THE DRIVE TRAIN ELECTRIFICATION FOR A RESCUE BOAT2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Progressing constraints on green house gas emissions lead to a sustainability trend,which greatly a↵ects the transport sector. Nowadays, companies show increasing interest indeveloping sustainable solutions.This thesis has been started thanks to a project given by Sj¨or¨addningss¨allskapet, themost relevant association that performs sea rescue operations in Swedish waters.Sj¨or¨addningss¨allskapet would like to explore the possibility of making their rescue boat fleetentirely carbon-free, hence more sustainable.What may provide a suitable solution is an electric drive train with hybrid energy storage,composed by a battery pack and a fuel cell stack. The research question is whether it wouldbe feasible to combine fuel cell stacks and battery packs to provide power to a fast smallboat.From a sketch of a rescue boat, the drive train design for such boat is studied in itsintegrity, from the water jet pump to the battery and fuel cell systems.The required power has been calculated empirically, using data from online tests on waterjet boats. Di↵erent tests have been considered, resulting in a mean power curve and a meanconsumption curve and allowing comparison between the hybrid electric drive train withan internal combustion engine drive train.Three profiles of speed, power and consumption have been assumed for the calculation ofthe required energy and hence rate the energy storage system. A design has been proposedin terms of fuel cell capacity and battery capacity.The propulsion unit, composed by the electric machine and water jet, has been studied,focusing on di↵erent electric drive technologies. Few conclusions on both the weight andsustainability requirements are discussed.A sustainability analysis is carried out in terms of CO2 emissions, through a life cycleassessment accounting for the environmental impact of the system during the whole lifecycle, from cradle to grave.

  • 34.
    Andung Muntaha, Muhamad
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Non-intrusive Logging and Monitoring System of a Parameterized Hardware-in-the-loop Real-Time Simulator2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Electronic Control Unit (ECU) is a crucial component in today’s vehicle. In a complete vehicle, there are many ECUs installed. Each of these controls a single function of the vehicle. During the development cycle of an ECU, its functionality needs to be validated against the requirement specification. The Hardware-in-the-loop (HIL) method is commonly used to do this by testing the ECU in a virtual representation of its controlled system. One crucial part of the HIL testing method is an intermediary component that acts as a bridge between the simulation computer and the ECU under test. This component runs a parameterized real-time system that translates messages from the simulation computer to the ECU under test and vice versa. It has a strict real-time requirement for each of its tasks to complete.A logging and monitoring system is needed to ensure that the intermediary component is functioning correctly. This functionality is implemented in the form of low priority additional tasks that run concurrently with the high priority message translation tasks. The implementation of these tasks, alongside with a distributed system to support the logging and monitoring functionality, is presented in this thesis work.Several execution time measurements are carried out to get the information on how the parameters of a task affect its execution time. Then, the linear regression analysis is used to model the execution time estimation of the parameterized tasks. Finally, the time demand analysis is utilized to provide a guarantee that the system is schedulable.

  • 35.
    Angeles Antolin Linan, Maria
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Effects of load modelling on Voltage Impasse Regions (VIR)2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Voltage Impasse Region (VIR) is a phenomenon in power systemswhose dynamics are describe by a set of Differential AlgebraicEquations (DAE). VIR denotes a state-space area where voltagecausality is lost, i.e. the Jacobian of the algebraic part of DAEis singular. In a Time Domain Simulation (TDS) once system trajectoriesenter VIR, TDS experiences non-convergence of the solution.Then, there is no reason to continue with the simulation. Thisis why it is important to understand the mechanisms that introduceVIR. It is known that VIR appears in relation to static, non-linearload models. However, it remained unknown what the cumulativeeffect of several static, non-linear loads would be.This master thesis has further expanded the concept of VIRby carrying out a structured study on how the load modelling affectsVIR. For this purpose, this thesis proposes a quasi-dynamicmethodology to map VIR in the relative rotor angle space. Themethodology introduces a new discrete index called Voltage ImpasseRegion Flag (VIRflag), which allows to determine if the algebraicequations of DAE are solvable or not and, thus, to locate VIR.A test system is used to test the proposed quasi-dynamic approach.The VIRflag was first used to map VIR for various load combinations.Then, the relationship between TDS non-convergence issuesand the intersection of a trajectory with VIR is examined toverify the proposed methodology.The proposed method has been proved to be efficient in the determinationof VIR regardless of the number of non-linear loads inthe power system. Among the static exponential load models, theConstant Power (CP) load component has been identified as theone with the largest influence on VIR appearance and shape. TheConstant Current (CC) loads induce ”smaller" VIR areas and theConstant Impedance (CI) load can only alter the shape of VIR inthe presence of non-linear load models.

  • 36.
    Anghileri, Davide
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Using Player Modeling to Improve Automatic Playtesting2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    In this thesis we present two approaches to improve automatic playtesting using player modeling. By modeling various cohorts of players we are able to train Convolutional Neural Network based agents that simulate human gameplay using different strategies directly learnt from real player data. The goal is to use the developed agents to predict useful metrics of newly created game content.

    We validated our approaches using the game Candy Crush Saga, a non-deterministic match-three puzzle game with a huge search space and more than three thousand levels available. To the best of our knowledge this is the first time that player modeling is applied in a match-three puzzle game. Nevertheless, the presented approaches are general and can be extended to other games as well. The proposed methods are compared to a baseline approach that simulates gameplay using a single strategy learnt from random gameplay data. Results show that by simulating different strategies, our approaches can more accurately predict the level difficulty, measured as the players’ success rate, on new levels. Both the approaches improved the mean absolute error by 13% and the mean squared error by approximately 23% when predicting with linear regression models. Furthermore, the proposed approaches can provide useful insights to better understand the players and the game.

  • 37.
    Anna, Canal Garcia
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Study of brain imaging correlates of Mild Cognitive Impairment and Alzheimer’s Disease with machine learning2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Accurate diagnosis in the early stages is an important challenge for the prevention and effective treatment of Alzheimer’s Disease (AD). This work proposes a method of analysis of the correlation of Mild Cognitive Impairment (MCI) subtypes and its progression to AD using neuroimages such as structural magnetic resonance imaging (MRI) scans. Basic data pre-processing such as the extraction of brain-tissue related parts of the image, image registration and standardization to the mean and deviation is applied. A convolutional autoencoder (CAE) is used to reduce data dimensionality and learn generic features capturing AD biomarkers, followed by various clustering techniques in order to detect different patterns on MCI data. In addition, six MCI patient clusters are generated based on AD progression information provided by ADNI. The method is evaluated on a total of 1069 structural MRI scans (522 MCI scans, 243 AD scans and, 304 CN scans) on the baseline from ADNI database. No clearly separable clusters are found after using CAE model trained on MCI data. Therefore, it is difficult to confirm a strong correlation between different subtypes of MCI patients and its progression to AD. Nevertheless, a significant correlation within the baseline images of the respective six groups identified based on AD progression is reported. It is hypothesized that lack of domain-specific MRI processing, planned in this work, could be a deciding factor about the findings in this research.

  • 38.
    Annergren, Björn
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Log Classification using a Shallow-and-Wide Convolutional Neural Network and Log Keys2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    A dataset consisting of logs describing results of tests from a single Build and Test process, used in a Continous Integration setting, is utilized to automate categorization of the logs according to failure types. Two different features are evaluated, words and log keys, using unordered document matrices as document representations to determine the viability of log keys. The experiment uses Multinomial Naive Bayes, MNB, classifiers and multi-class Support Vector Machines, SVM, to establish the performance of the different features. The experiment indicates that log keys are equivalent to using words whilst achieving a great reduction in dictionary size. Three different multi-layer perceptrons are evaluated on the log key document matrices achieving slightly higher cross-validation accuracies than the SVM. A shallow-and-wide Convolutional Neural Network, CNN, is then designed using temporal sequences of log keys as document representations. The top performing model of each model architecture is evaluated on a test set except for the MNB classifiers as the MNB had subpar performance during cross-validation. The test set evaluation indicates that the CNN is superior to the other models.

  • 39.
    Anwar, Monib
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Effect of X-ray Irradiation on the Blinking of CdSe/ZnS Nanocrystals2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Different semiconductor nanocrystals exhibit size dependent properties due to confinement effect. Light emission from these nanocrystals may turn ON and OFF seemingly at random, an effect known as blinking. In this work blinking studies have been done to monitor the effect of X-ray exposure and to investigate the radiation hardness of CdSe/ZnS QD’s. Correct parameters to dilute and spin-coat the obtained sample were found to get access to individual single dots. Blinking of these dots was analyzed using Image J and MATLAB plug-in, where ON and OFF-times distribution power exponents Mon and Moff have been extracted to see the change in emission intermittency after a total cumulative dose of ~1026 Gy (absorbed by SiO2) in steps. It was observed that blinking was quenched and consequently the QD’s went permanently to off state as a result of X-ray exposure.

  • 40.
    Arcidiacono, Claudio Salvatore
    KTH, School of Electrical Engineering and Computer Science (EECS).
    An empirical study on synthetic image generation techniques for object detectors2018Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Convolutional Neural Networks are a very powerful machine learning tool that outperformed other techniques in image recognition tasks. The biggest drawback of this method is the massive amount of training data required, since producing training data for image recognition tasks is very labor intensive. To tackle this issue, different techniques have been proposed to generate synthetic training data automatically. These synthetic data generation techniques can be grouped in two categories: the first category generates synthetic images using computer graphic software and CAD models of the objects to recognize; the second category generates synthetic images by cutting the object from an image and pasting it on another image. Since both techniques have their pros and cons, it would be interesting for industries to investigate more in depth the two approaches. A common use case in industrial scenarios is detecting and classifying objects inside an image. Different objects appertaining to classes relevant in industrial scenarios are often undistinguishable (for example, they all the same component). For these reasons, this thesis work aims to answer the research question “Among the CAD model generation techniques, the Cut-paste generation techniques and a combination of the two techniques, which technique is more suitable for generating images for training object detectors in industrial scenarios”. In order to answer the research question, two synthetic image generation techniques appertaining to the two categories are proposed.The proposed techniques are tailored for applications where all the objects appertaining to the same class are indistinguishable, but they can also be extended to other applications. The two synthetic image generation techniques are compared measuring the performances of an object detector trained using synthetic images on a test dataset of real images. The performances of the two synthetic data generation techniques used for data augmentation have been also measured. The empirical results show that the CAD models generation technique works significantly better than the Cut-Paste generation technique where synthetic images are the only source of training data (61% better),whereas the two generation techniques perform equally good as data augmentation techniques. Moreover, the empirical results show that the models trained using only synthetic images performs almost as good as the model trained using real images (7,4% worse) and that augmenting the dataset of real images using synthetic images improves the performances of the model (9,5% better).

  • 41.
    Ardal, Dui
    KTH, School of Electrical Engineering and Computer Science (EECS).
    A Collaborative Previsualization Tool for Filmmaking in Virtual Reality2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Previsualization is a process within pre-production of filmmaking where filmmakers can visually plan specific scenes with camera works, lighting, character movements, etc. We developed and assess a prototype for previsualization in virtual reality for collaborative purposes where multiple filmmakers can be present in a virtual environment to share a creative work experience, remotely. The costs of computer graphics-based effects are substantial within film production, using previsualization, these scenes can be planned in detail to reduce the amount of work put on effects in the later production phase. By performing a within-group study on 20 filmmakers, our findings show that the use of virtual reality for distributed, collaborative previsualization processes is useful for real-life pre-production purposes. These results provide insights on how to best design collaborative, virtual reality-applications used for remote work, and highlights the pitfalls of certain design choices.

  • 42.
    Ardila Bernal, Pablo Andres
    KTH, School of Electrical Engineering and Computer Science (EECS).