3456 251 - 284 of 284
rss atomLink to result list
Permanent link
Cite
Citation style
  • apa
  • harvard1
  • 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
  • Kam, Felix
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Mellin, Henrik
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Different frequencies of maneuver replanning on autonomous vehicles2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    In recent years has autonomous vehicles (AV’s) become a more and more relevant topic. Several big car companies race to get their model out on the market as soon as possible. However building a fully functional AV is not an easy task. One of the main reasons behind this is that safely operating a vehicle requires advanced decision-making. Such advanced decision-making requires a lot of computational power which can lead to the AV taking long time to reach a decision. In this thesis explored how long that time could be while at the same time having the AV drive safely. To do this we created a simulation with an AV alongside randomly generated traffic and limited how often it would make new decisions. Our results showed that updating every 0.5s was needed for driving safely at 50, 70 and 90 km/h and 0.3s was needed for 110 km/h.

  • Hansson, Hanna
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Zetterlund, Louise
    KTH, School of Electrical Engineering and Computer Science (EECS).
    A study of cooperation between humans and robots in warehouse intersections2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    In the modern world, many warehouses use robots to collect different orders and to keep humans safe the robots work in restricted areas. This practice could be seen as a waste of space and resources. Therefore, this paper sets out to investigate whether it is time efficient or not. The study clocked 27 test subjects completing a simple task in a simulated environment alongside a robot, and compared the time against two robots completing the same tasks. The timer stopped when both parties had completed their task. The simulated environment consisted of an intersection and a script making the robot give way to the human-navigated robot. In the control environment with the two robots, hierarchical protocols were used, making the robot with less work left to give way to the other. The results of the study showed that it is more time efficient with only robots working in this simulated environment, compared to having both a human and a robot sharing work area. However, the results also showed that the human approach towards the robot affected the result. A test subject with a positive view on cooperating with the robot resulted in a shorter time spent on completing the task than test subjects not taking the robot into account. This could imply that it is possible to make the human-robot interaction more efficient by altering the human approach towards the robot.

  • Bergendal, Rasmus
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Rohlén, Andreas
    KTH, School of Electrical Engineering and Computer Science (EECS).
    A comparison of training algorithms when training a Convolutional Neural Network for classifying road signs2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    This thesis is a comparison between three dierent training algorithms when training a Convolutional Neural Network for classifying road signs. The algorithms that were compared were Gradient Descent, Adadelta, and Adam. For this study the German Trac Sign Recognition Benchmark (GTSRB) was used, which is a scientically relevant dataset containing around 50000 annotated images. A combination of supervised and offline learning was used and the top accuracy of each algorithm was registered. Adam achieved the highest accuracy, followed by Adadelta and then GradientDescent. Improvements to the neural network were implemented in form of more convolutional layers and more feature recognizing filters. This improved the accuracy of the CNN trained with Adam by 0.76 percentagepoints

  • Ardö, Edvin
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Lindholm, Johan
    KTH, School of Electrical Engineering and Computer Science (EECS).
    A comparative study between a genetic algorithm and a simulated annealing algorithm for solving the order batching problem2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Optimizing warehouse automation requires finding efficient routes for pickingup items. Dividing the orders into batches is a realistic requirement for warehouses to have. This problem, known as the order batching problem, is an NP-hard problem. This thesis implements and compares two meta-heuristics to the order batching problem, simulated annealing (SA) and a genetic algorithm(GA). SA was found to perform equal to or better than GA on all occasions in terms of minimizing traveling distance. The algorithms were tested on 6 different warehouses with various layouts. The algorithms performed similarly on the smallest problem size, but in the largest problem size SA managed to find 17.1 % shorter solutions than GA. SA tended to find shorter solutions in a smaller amount of time as well.

  • Nyqvist, Anna
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Rutqvist, Joar
    KTH, School of Electrical Engineering and Computer Science (EECS).
    The Impact of Colour Themes on Code Readability2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    The ability to read code is an important skill in programming, and is said to be affected by the coding environment and interface. This thesis investigates the impact of colour themes on code readability, comparing dark text on a light background to light text on a dark background. An experiment was performed where participants solved code comprehension tasks while recorded using eye tracking. Results indicated no significant differences in accuracy or time between the two colour themes. Similarly, the eye tracker recordings showed no significant difference in eye movement patterns between colour themes.

  • Norrman, David
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Vilhelmsson, Isak
    KTH, School of Electrical Engineering and Computer Science (EECS).
    The Impact of Improper Indentation on Code Comprehensibility2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    This study uses eye-tracking and time measuring to examine if improperly indented code affects code comprehension in Java source code. The study was divided into three different phases of testing, each with three samples of Java source code with various degrees of indentation. During the tests, the test subjects were asked to read the code and try to determine what the output would be. The tests had a total of 33 test subjects. Results from the time measuring and the amount of correctly determined outputs implied a correlation between a proper indentation and how easy the code was to comprehend, where proper indentation made it easier. The eye-tracking results did not give any conclusive answers to the question.

  • Henning, Johan
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Hellesnes, Nicolai
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Detecting Plagiarism Patterns in student code2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Plagiarism has become a big concern in programming both in education and in the industry of software development. While a lot of effort have been put into detecting plagiarism, most of the it have been focused on detecting plagiarism in plain text. The methods for cheating has evolved as plagiarism detection has improved. This thesis looks at plagiarism in entry level programming courses to discover how wide spread the cheating is, and if plagiarism detection algorithms in conjunction with metadata from GitHub can be used to better detect cheating. More specifically the commit metadata from GitHub is used to see if any interesting patterns with students who plagiarize can be found. The dataset used in this thesis are GitHub repositories for the entry level programming courses DD1337 and DD1338 for the year of 2015. The data set consists of 17 programming assignments with around 200 student submissions per assignment. The plagiarism detection tools used were MOSS and for each week the 10 most suspicious submitted assignments were added to a suspicious-list which were later used to help find patterns in students that plagiarize. The results show that the suspicious students on average had 5.27 commits per assignment, while the non-suspicious students had 6.49 commits on average per assignment. This is to say that suspicious students on average had a lower number of commits than the non-suspicious students. Future work includes testing with bigger data sets, and testing other metadata for finding other interesting patterns in cases of plagiarism.

  • Closa Márquez, Oriol
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Decyphering the Geheimschreiber, a Machine Learning approach: Recreating and breaking the Siemens and Halske T52 used during World War II to secure communications in Sweden2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Historically, rotor cyphers have been used in order to secure written communications. Mechanical machines provided continuous streams of characters for encoding secret messages that were sent to the other part of the continent by means of telephone cables or radio. Several people tried in vain to tackle them but only those bold enough were successful. In Sweden, the Siemensand Halske T52 was used by the Germans during World War II and Arne Beurling was one of those bright people that successfully broke it. This thesis aims to recreate his steps applying modern concepts to the task, breaking the Geheimschreiber. In order to do that, a recreation of the machine has virtually been built and several German texts encyphered. The techniques used, involving Recurrent Neural Networks, have proven to be effective in breaking all XOR wheels with different crib sizes removing the random factor introduced by the cypher. However, if this method can be applied to real war intercepts remains to be seen.

  • Broberg, Pontus
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Jahanshahi, Shapour
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Using eye tracking to study variable naming conventions and their effect on code readability2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Using camel case when naming variables is largely considered to be best practise when writing code these days. But is it really the best variable naming convention when it comes to code readability and understanding? And how does different variable naming conventions affect the readability of code? This thesis researches these questions using eye tracking technology. Test subjects are timed as they look at and explain code snippets using different variable naming conventions while their gaze is plotted onto a heatmap. The variable naming conventions tested were single letters, single words, multiple words in camel case and multiple words in snake case. From the results shown, the conclusion is drawn that no significant difference in readability can be confirmed between the different variable naming conventions.

  • Bakhuizen, Ellinor
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Landelius, Cecilia
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Diverge - Investigating the Consequences of Bad Comments2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    In large software development projects the majority of code comments are written at the beginning of the project and tend to not be updated when the code is rewritten. This commonly results in code with incorrect comments or no comments at all. This study intends to answer whether incorrect comments can mislead programmers and if well written comments assist programmers ininterpreting new code. Furthermore, the attitudes towards code comments are investigated. The research questions were answered with data from forms andc ode tests run on 35 engineering students. For the code tests eye trackers were used to provide a clear picture of how much the participants read the code and the associated comments. A majority of the test subjects agreed that comments are important whilst 8% considered comments to be unnecessary. 50% of the test subjects expressed positive feelings towards writing comments. The data from the eyetracker showed that the test subjects read comments and code equally. Thes tudy found that incorrect comments in many cases lead to misconceptions. Correct comments were shown to assist the programmer if the code contained library functions that the programmer was not familiar with. Regarding correct comments versus no comments at all there was no difference in readability if the code did not contain any library functions.

  • Westerberg, Adrian
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Ung, Gustav
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Comparing Verification of List Functions in LiquidHaskell and Idris2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Formal verification is important for guaranteeing correctness of critical computer programs. LiquidHaskell and Idris are 2 functional programming languages in which formal verification can be performed using the respective type systems. This thesis investigates the differences between LiquidHaskell and Idris when verifying list functions, with respect to expressiveness, annotation overhead and type checking times. The used list functions, as well as their specifications consisting of correctness properties, were carefully selected. No large differences in expressiveness were found between LiquidHaskell and Idris. However, Idris was found to be better than LiquidHaskell without the automation feature PLE, regarding annotation overhead and type checking times. With PLE enabled, no large differences were found.

  • Isabar, Diyala
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Aurell Hansson, Björn
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Preventing Interference in Smart Contracts2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Smart contracts are a new form of contracts that handle transactions of economical value between users on a network, without the need for a middle man. This is possible as the integrity of the transactions are ensured using Blockchain technology. However, safe transactions are not always a guarantee as secure smart contracts can be difficult to write. Security vulnerabilities in smart contracts can be exploited and result in great financial losses for contract users.

    In this report we examine how interference can be counteracted in transactions involving smart contracts. This is to ensure that assets of economical value do not run the risk of disappearing and that the flow of information between smart contracts remains truthful. The term "interference" is coined precisely in relation to smart contracts. Moreover, two different solutions for two interference problems are examined, which mainly involve locking the access to certain functions for specific contracts so that no modification of the state of one contract can take place until a transaction is finished. The methods for counteracting interference in the transactions work, but the solutions are adapted to the specific pr blems and do not serve as a "general" solution to the problem of interference.

  • Handberg, Erik
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Rostami, Lara
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Epistemic Structures for Strategic Reasoning in Multi-Player Games2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    A game can be visualised with a directed graph, where each node is a game state and the edges are the players’ actions that lead to new game states. To obtain the winning condition of the game whilst avoiding the losing condition, the players must perform the correct actions in coalition. The goal is thus to find winning strategies for these games, which ensure that the winning condition is reached. For games with imperfect information, the players might not be able to distinguish some game states from each other, which makes finding strategies harder. With the multiplayer knowledge subset construction (MKBSC), one may reduce the original graph to an “expanded” graph, and from this new graph try to find winning strategies, that can later be translated back into the original game graph. In this report, we have investigated how we gain information in each iteration of the MKBSC. This was done by introducing e-trees, a mathematical tree construction that can visualise knowledge of the current game state for each player in the game. It was found that these e-trees can display how the knowledge develops for each player given any iteration of the MKBSC, and that they can provide an intuitive way of understanding how strategies are found in a game.

  • Amilon, Jesper
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Lindberg, Axel
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Verifying Temporal Properties Using Deductive Verifiers2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Formal verification is an area of theoretical computer science where mathematical logic is used to prove that a program behaves in a certain way. With the methods in formal verification, you can prove that the program follows some given specification and thereby behaves in the desired way. The area is largely split up into two distinct parts. One deals with how the program transforms data. This uses Hoare logic and deductive verification to prove that the program follows a given specification. The other part deals with temporal properties of the program, this uses temporal logic and model checkers.

    The two areas are today largely separated. This report builds on a framework by Alur and Chaudhuri [1] which proves temporal properties in a Hoare logic style reasoning. By using this framework, this report aims to check the viability of using it with deductive verifiers. Thereby bridging the gap between the two areas for formal verification.

    In conclusion, the report finds that it is certainly possible to prove temporal properties for C programs using Alur and Chaudhuri’s framework with deductive verifiers. In practical terms, though, it requires too much work to be feasible to use this framework by manually creating annotations for the deductive verifiers. In a small example program of 13 lines, proving a temporal property required around 50 extra lines of annotations. However, some parts of the annotation process could be automated with tooling support but to achieve full automation is probably not possible. This is partly due to ranking functions that the framework requires which, in general, are not easy to generate automatically.

  • Sunnegårdh, Christina
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Eserstam, Klara
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Intention-revealing function names and small functions to facilitate code comprehension2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Code comprehension is by many considered to be one of the most expensive and time-consuming phases of the software life cycle. There are multiple techniques for making code more comprehensible, one of them alleged to be keeping functions small. However, the claim that small functions are superior to large functions with regards to code comprehension, are often based upon programming experience and stated without references from research. In these claims the importance of intention-revealing function names to improve codecomprehension is also emphasized. It could therefore be questioned whetherit is keeping functions small or the enabling of intention-revealing functionn ames that underlies the claim of superiority of small functions. This thesis presents previous studies and relevant literature within the area as well as carries out tests using eye tracking. The results of the tests indicated that intention-revealing function names have a significant effect on how fast developers comprehend code. It is also indicated that simply splitting up a function, without using intention-revealing function names, increases the chance for programmers to be able to tell the correct output. To be able to draw reliable conclusions, further studies would be necessary, where the most important improvements would be to provide an unlimited amount of time for each test as well as a larger quantity of test persons.

  • Nilsson, Martin
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Ekström, Erik
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Investigating the Epistemic Unfolding of Multi-Player Games of Imperfect Information2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Solving multi-agent planning problems entails finding strategies that ensure that the agents achieve their cooperative objective. In practice, since the agents do not necessarily have perfect knowledge, these planning problems may involve imperfect information. For example, a robot agent may have a damaged or imperfect sensor, or one agent may be in another room and therefore cannot see what the other agents can. A useful model for these problems are cooperative multi-player games of imperfect information. In this paper, we investigate a novel construction known as epistemic unfolding. Epistemic unfolding translates multi-player games of imperfect information to single-player games of perfect information, for which strategy synthesis is much easier. Essentially, the idea of epistemic unfolding is to track the knowledge of the players throughout the duration of the game. Although epistemic unfolding is previously defined, to our knowledge it had neither been implemented nor properly visualized. In order to provide a foundation for future research, we implemented epistemic unfolding as a tool, visualized a set of epistemically unfolded games, and, finally, developed an intuitive understanding of epistemically unfolded games.

  • Mitic, Aleksander
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Nehlin, Oskar
    KTH, School of Electrical Engineering and Computer Science (EECS).
    A Comparison of Different Machine Learning Models for Cardiovascular Disease Detection2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Cardiovascular disease (CVD) is the leading cause of death worldwide and the majority of the deaths occur in low to middle income countries. This makes the prevention of CVDs an accute problem to study and much research has been done already. New techniques using computer aided diagnostics with the help of machine learning (ML) might have great potential in the future of diagnostics of CVDs but the area of research is relatively new. In this report we compare three distinct models of machine learning, Support Vector Machines (SVM), Artificial Neural Networks (ANN) and Decision Trees (DT). We compare their current classification accuracy and potential viability in CVD classification. Ensemble learning and other model specific optimizations were out of scope for this report and a more general and basic implementation was used. Our results does not indicate a clear winner and all models have different pros and cons. The average accuracy did not differ much between the different models. We found that the SVM gave the highest average classification accuracy while the ANN had similar peek classification accuracy however a slightly lower average classification accuracy. The DT however gave the most interpretable results since the trained model can be easily visualized which made us conclude that DTs are today perhaps the most viable option to be used, as a complement to physicians in their current methods of diagnostics. These report results are although limited by the dataset which was too small for any wide stretching and general conclusions.

  • Yamani, Zuhoor
    KTH, School of Engineering Sciences in Chemistry, Biotechnology and Health (CBH).
    Design of Multi-function Polymeric Nanoparticles for Theranostic Application2019Independent thesis Advanced level (degree of Master (Two Years)), 80 credits / 120 HE creditsStudent thesis
    Abstract [en]

    Block copolymer nanoparticles (NPs) have gained great attention among researcher for various medical application mainly due to their extraordinary optical, chemical, and biological properties. The current thesis presents design of multifunctional polymeric NPs for imaging and drug delivery system (DDS) with an in-vitro study of their participation in drug release and cell viability. The NPs were synthesized using reversible addition chain fragmentation transfer (RAFT)-mediated emulsion polymerization via polymerization induce self-assembly (PISA) approach. The environment-friendly emulsion polymerization process of n-buytl acrylate (n-BA) in water is highly efficient. The process produced uniform NPs which would have control over the particle size and molecular weight of the compound. Herein we report a novel simultaneous encapsulation of camptothecin (CPT) and Nile red (NR) into poly(ethylene glycol) methyl ether methacrylate-co-N-hydroxyethyl acrylamide-b-poly n-buytlacrylate (PEGA-co-HEAA)-b-P(n-BA) during the particles formation with a small particle size of 66 nm, high conversion ~80% and encapsulation efficiency of ~50%. The In vitro drug release of the CPT from the NPs exhibited an initial burst (70-80%) within 6h. cell viability was evaluated for the NPs against RAW 264.7 cell line, which indicated the designed NPs are biocompatible and not toxic.

     

  • Lundstig, Karl
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Evaluating different training techniques for a convolutional neural network that classifies Alzheimer’s disease2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Effective computer diagnosis of Alzheimer’s disease could bring large benefitsto the millions of people worldwide who does or will suffer from dementia. One popular method for trying to achieve this is the training of convolutional neural networks to classify MRI brain scans. An abundance of training methods exists that aims to improve the performance of these neural networks, but their effectiveness and eventual disadvantages are not always clear. This study evaluates the performance of a CNN on the OASIS-3 neuroimaging dataset. The CNN’s task is to classify each MRI scan into one of four classes which correspond to the stage of Alzheimer’s disease. Two different training methods are compared to a training baseline. The first method evaluated is class-weighting, a technique that tries to compensate for rare classes in imbalanced datasets. The second method is data augmentation, a techniquet hat extends the dataset in an attempt to reduce overfitting and increase performance. Class-weighting was found to improve the classification performance significantly on the rarest class, while not having too large effect on the other classes. Data augmentation was not found to improve performance in general, but did improve the recall on some classes.

  • Andersson, Olle
    KTH, School of Engineering Sciences in Chemistry, Biotechnology and Health (CBH).
    Predicting Patient Length Of Stay at Time of Admission Using Machine Learning2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    This master thesis investigates the possibility of using machine learning methods to predict patient length of stay at the time of admission to a clinical ward from the emergency department. The main aim of this thesis is to provide a comparative analysis of different algorithms and to suggest a suitable model that can be used in a hospital prediction software. The results show that it is possible to achieve a balanced accuracy of 0.72 at the time of admission and of 0.75 at a later stage in the process. The suggested algorithm was Random Forest which combines good accuracy with effective training time, making it suitable for on-line use in a hospital. The study shows that there is a clear potential for the use of machine learning methods for predicting length of stay, but that further improvements have to be made before adaption into the healthcare.

  • Lindqvist, Björn
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Reproducing the state of the art in onset detection using neural networks2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Great strides have been made in the state of the art performance of musicial onset detection in recent years with better and better detectors being invented at a fast pace. The current top spot is held by Schlüter and Böck, who in 2014 presented a detector based on a convolutional neural network (CNN) that attained an F-score of 90.3% (Precision 91.7%, 88.9% recall) on a commonly used dataset [1].

    In 2018 two researchers, Gong and Serra, tried to replicate their result but only reached an F-score of 86.67% (precision and recall values weren’t reported) [2], a significantly worse result than Schlüter and Böck’s. In comparison a 2013 detector based on a recurrent network, also designed by Schlüter and Böck, achieved an F-score of 87.3% [3].

    Gong and Serra’s result casts doubt on the 90.3% figure reported by Schlüter and Böck. We therefore try to shed some light on the question of what the state of the art performance in musical onset detection is by posing and answering the question; can Schlüter and Böck’s result be reproduced?

    Our answer is “Maybe – but we were unable to!” which is perhaps the only result possible since you can’t prove a negative. We trained the CNN architecture three times and obtained F-scores of 85.0%, 85.8% and 85.6%. For the RNN architecture, which we also tried to reproduce, we obtained the scores 86.3%, 86.3% and 86.3%. Due to omission of details that perhaps were significant from the referenced articles, we weren’t able to recreate Schlüter and Böck’s architectures exactly and had to make some “educated guesses.” It is possible that those guesses caused performance to suffer. Nevertheless, we believe that our work is worthwhile because it demonstrates how infuriatingly difficult it is in deep learning for researchers to reproduce each others work.

  • Lindblad, Maria
    KTH, School of Electrical Engineering and Computer Science (EECS).
    A Comparison of a Heuristic and a Hopfield Neural Network Approach for Solving Examination Timetabling Problems2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    The Examination Timetabling Problem (ETP) is the problem of scheduling a number of exams during a set time period so that no students are required to sit two exams simultaneously. Despite the complexity of the problem, universities all over the world solve ETPs several times each year. Two known methods for solving ETPs is using either heuristics or Hopfield Neural Networks (HNN). This thesis compares the performance of a heuristic algorithm implemented with Local Search, Simulated Annealing and Tabu Search to the performance of a HNN algorithm. Both algorithms were executed on ten different ETPs reduced to Graph Colouring Problems (GCP). The results show that the heuristic algorithm always generated more satisfactory solutions to the ETPs than the HNN. The HNN was, however, implemented as software in this thesis. It is intended to be implemented as hardware and if this method were to have been used instead the HNN algorithm might have produced other results. At this stage the heuristic algorithm is more suitable than the HNN algorithm for solving ETPs.

  • Kolanowski, Mikael
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Stevens, David
    KTH, School of Electrical Engineering and Computer Science (EECS).
    A Comparative Study of the Effect of Features on Neural Networks within Computer-Aided Diagnosis of Alzheimer's Disease2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Alzheimer’s disease is a neurodegenerative disease that affects approximately 6% of the global population aged over 65 and is forecasted to become even more prevalent in the future. Accurately diagnosing the disease in an early stage can play a large role in improving the quality of life for the patient. One key development for performing this diagnosis is applying machine learning to perform computer-aided diagnosis. Current research in the field has been focused on removing assumptions about the used data sets, but in doing so they have often discarded objective metadata such as the patient’s age, sex or priormedical history. This study aimed to investigate the effect of including such metadata as additional input features to neural networks used for diagnosing Alzheimer’s disease through binary classification of magnetic resonance imaging scans. Two similar neural networks were developed and compared, one with these additional features and the other without them. Including the metadata led to significant improvements in the network’s classification accuracy, and should therefore be considered in future computer-aided diagnostic systems for Alzheimer’s disease.

  • Häregård, Erik
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Kruger, Alexander
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Comparing syntax highlightings and their effects on code comprehension2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Syntax highlight is a system designed to assist a writer or programmer by displayingdifferent parts of a text in a specific color based on its function. In this study, we conducted a practical experiment comparing the effectiveness of two common syntax highlightings by primarily measuring the speed of which participants could understand a given piece of code. The highlights chosen forthe comparison were Atom UI Standard and Styri. Previous studies in this area, most notably by Advait Sarkar from the University of Cambridge, have shown a generally positive effect of syntax highlight but have not compared how different types of syntax highlightings affected the reader. Our study was performed with eight participants from a Technical High school with a minor background in programming who answered six questions where some code was highlighted and some not; the order of the items was found to affect results on some questions. The results do not show any significant advantage to using syntax highlight or a difference between the effectiveness of the two syntax highlights. Some participants saw a slight constant advantage from using syntax highlight, but no general or notable conclusion could be made about these participants.

  • Snygg, Oliver
    et al.
    KTH, School of Industrial Engineering and Management (ITM), Production Engineering.
    Torberger, Carl
    KTH, School of Industrial Engineering and Management (ITM), Production Engineering.
    Vilka nya markanadsmodeller kanfå producerande företag att gå motcirkulär ekonomi?2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Environmental awareness in the world is increasing. The social debate is increasingly formed by an environmental perspective on several fronts. More and more people are beginning to understand that the earth's resources are finite and we must think about how we live today. The producing industry accounts for a large part of the use of raw materials, but how will the industry change in the future?

    The industry of the future is often described as a circular economy where minimal amounts of raw materials are broken and the majority of the material that is in use goes back to production. The reality we live in today is another, the greater part of the world's industry continues with linear production where raw materials are used in manufacturing and products are deposited at the end of its lifecycle. The project group examines the building blocks of circular economy and where we are today in the development towards sustainable management of raw materials.

    The purpose of the report is to investigate the economic advantages for companies to change to circular economy in the form of new business models. Opportunities for these business models often grow with the help of new technology from Industry 4.0 like Internet ofThings and Big Data.

    We have the conclusion that business models such as pay-per-use where companies maintain ownership of the product, often are easy to implement in circular economics, since the company itself can control the entire circuit for the product. They make it easier to base their production on return of products if they control the volumes of the returned products themself. These models are often based on selling services that are dependent on the Internet of things and Big Data.

  • Heddini, August
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Wihl, Konrad
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Strategy Synthesis for Multi-Agent Systems with Imperfect Information and Imperfect Recall2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    This report aims to study and compare currently available tools for uniform strategy synthesis in multi-player versus environment games of imperfect information and imperfect recall. These games are represented as a graph of game states, and goal sequences of states are typically represented using temporal logic formulae. Strategies for achieving those goals can then be found by evaluating the formulae. Specifically the time complexity and addressed game objectives of the tools were of interest. Not too much is known about this particular field of game theory, and strategy synthesis is complicated and computationally heavy. Today there is only one completed tool specialized for the particular game structure of interest, the Strategic Model Checker (SMC). Three tools were found which could be used for the task, out of which two seemed worthy of further investigation. Firstly, the aforementioned Strategic Model Checker (SMC) provided a good reference as a completed program. Secondly, an experimental version of the ATLir Model Checker, a newer tool still under development specialized for the same task, could provide an example at the cutting edge of the field. The evaluation of the literature as well as the tools confirmed and corroborated earlier findings in that there are few implementations available, that the time complexity of their performance is steep, albeit improving, and that they deal with a potentially very wide range of game objectives, which can be described using Alternating Time Temporal Logic, or ATL. The performance analysis concluded that the newer tool shows promising results for smaller games, but that SMC still produces more reliable results for larger instances as well as more complicated test cases.

  • Balcázar Castell, Iris
    KTH, School of Electrical Engineering and Computer Science (EECS).
    Markov chain Monte Carlo for the reconstruction of lineage trees from single-cell DNA data2019Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    The purpose of this study is to infer evolutionary trees through the Markovchain Monte Carlo algorithm (MCMC) [1] based on whole-genome single cell DNA sequencing data. By using MCMC we obtain likely tree structure samples according to the cells’ somatic point mutations in our data. This probabilistic framework takes into consideration the errors caused by the current technology such as amplification errors, sequencing errors and allelic dropouts. We investigated whether using this technique is reasonable given this biological scope. Most of the results give interesting conclusions that improve the previous results on the same Site Pair Model [2] and therefore we conclude that using MCMC is reasonable. Though, since the model is based on probabilities and the algorithm randomizes decisions the best results are not always guaranteed. One needs to be aware that a decent amount of data in the dataset is an important requisite to predict accurate tree structures. Furthermore, the computational time for this process is significantly high and can not be computed on regular laptops for large and realistic data sets. This is acceptable since for this type of research speed is not a strict requirement: it is worth waiting more for a given execution if the obtained results are more interestingor more accurate. Finally, we propose some further improvements for this strategy that could potentially obtain even better results in terms of accuracy and speed.

  • Gunnarsdóttir, Aðalheiður
    KTH, School of Engineering Sciences in Chemistry, Biotechnology and Health (CBH).
    Evaluation of Test Methods for Football Helmets Using Finite Element Simulations2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Introduction: Concussions in American Football are of a major concern due to highly reported injury rates. The importance of properly designed helmets have shown effect in reducing the risk of injuries, such as skull fractures. However, they are not as effective in reducing the risk of concussion. Helmets designed are required to pass standards and regulations for them to be allowed within the football leagues. The current test methods evaluate linear impacts, but lack evaluations of oblique impacts which are believed to cause concussions. Several test methods have been suggested, but little is known regarding how they compare.

    Objective: The purpose of this study was to compare three different test methods for evaluating helmet performance, utilizing finite element simulation. Three different helmet models were used for comparison, evaluating head kinematics. The helmet models were additionally ranked from best to worst based on their performances.

    Method: Three test methods, linear impactor, 45° angled linear impactor, and a drop test onto a 45° angled plate were simulated with three different open source helmet models. Simulations were conducted with one impact velocity at three impact locations. The influence of the interaction between helmet and head was also evaluated by altering the friction coefficient.

    Results: The test methods showed different results depending on helmet models, impact locations, and kinematics evaluated. Similarly, rankings of the helmets were varied based on methods and impact location. Little difference was observed after lowering the friction coefficient in majority of cases. The linear and angular acceleration for the drop side impact were mostly affected.

    Conclusion: Further evaluations of the test methods and comparison to real impacts is required to evaluate what method resembles head impacts best. Lowered friction coefficient had an effect for the drop impacts, but minor effect for other test methods

  • Berntsen, Jens
    et al.
    KTH, School of Industrial Engineering and Management (ITM), Learning.
    Bäckström, Jonathan
    KTH, School of Industrial Engineering and Management (ITM), Learning.
    Viktiga komponenter i e-lektioner: Implementation och evaluering av redigeringsverktyg för att skapa e-lektioner2019Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Throughout the ages, major technical breakthroughs have on several occasions been integrated into education, but each new technology has its strengths and weaknesses. With the development of the computer and its increased accessibility to the public, the amount of available teaching material (Clark & Mayer, 2016) and the number of students who are studying courses in e-learning has increased (Statista, 2015). E-learning is then, according to Clark & Mayer’s (2016) definition, learning that takes place through a digital medium. One possibility that arises with e-learning is that there are many ways of presenting material. One problem that arises then is to know which components should be used to create good e-lessons that promote student learning. The aim of the study is to identify which components are important for content creators to use in e-lessons and then, based on these components, develop an editing tool designed to create new e-lessons. To identify the appropriate components, a thematic analysis was first carried out on e-lessons from a learning platform in order to examine which components the instructional designers use to present material. Qualitative interviews were conducted to capture the instructional designers’ ideas and opinions regarding the components. These ideas and opinions were used together with the results of the thematic analysis to make didactic choices about which components are suitable for use in e-lessons. The development of the editing tool was done by creating multiple prototypes and discussing them with the instructional designers to determine the appearance of the components in the tool. From the thematic analysis, 7 themes emerged with a total of 22 sub-themes. The results of the didactic choices were 12 main components, for instance: text, image, video, quiz and examples. The editing tool that was built contains all 12 main components, but some components still lack some parts described in the didactic choices. The tool was evaluated by instructional designers who thought it worked but had some shortcomings that needs to be addressed.For the instructional designer to be able to use the editing tool to create and publish new e-lessons on the platform, the tool needs to be integrated into the platform, all components need to be completed and user experience improved.

  • Altayy, Fares
    et al.
    KTH, School of Industrial Engineering and Management (ITM), Learning.
    Schmied, Fredrik
    KTH, School of Industrial Engineering and Management (ITM), Learning.
    Stimulering av kritiskt tänkande och kollaboration i en programmeringsorienterad fysik- och matematiklaboration: En fallstudie med studenter från programmet Teknisktbasår (KTH)2019Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    This work aims to develop a proposal for educational material in the form of a programming-oriented Physics and Mathematics lab. The target group is mainly students from the second semester in the Technical Preparatory Year program (KTH), but also students of equivalent knowledge level (for example, students in the upper secondary school- natural science program). The lab was designed to create opportunity for students to:1) apply critical thinking, 2) demonstrate collaboration, 3) increase subject understanding. The proposed educational material comprises computational physics, where a non linear differential equation derived from a mathematical pendulum is solved using numerical methods. The exercises in the lab were formulated with critical thinking in mind and pair programming was used as the context for collaboration. In this paper 1) and 2) were investigated in an educational setting with students from the second semester of the Technical Preporatory Year (KTH). The conversations of four student groups were recorded and a content analysis was performed on the transcriptions of the audio recordings. A pattern matching method was used to strengthen the conclusions. Based on the results of the research project, we found that the theoretical proposition that the construction of the laboratory work has contributed to the collaboration and application of critical thinking by the students,has been strengthened.

  • Hellgren, Rikard
    et al.
    WSP Bridge and Hydraulic Design.
    Gasch, Tobias
    KTH, School of Architecture and the Built Environment (ABE), Civil and Architectural Engineering, Concrete Structures. Vattenfall Engineering, Stockholm, Sweden.
    Influence of fluid structure interaction on a concrete dam during seismic excitation2015Conference paper (Refereed)
    Abstract [en]

    The aim of this study is to investigate how Fluid-Structure interaction may be included in numerical earthquake analyses of dams. The basis for this project is theme A from ICOLDs 12th international benchmark workshop on numerical analysis of dams, which was held in October 2013. The focus of theme A was on how to account for fluid structure interaction in numerical earthquake analyses of dams. In this study, parametric numerical analyses have been performed where the purpose was to isolate some important parameters and investigate how these influence the results in seismic analyses of dams. These analyses were performed through the use of the finite element methodare the choice of Rayleigh damping parameters, reservoir boundaries and wave absorption in the foundation-reservoir interface. The use of acoustic elements has proven to be a powerful approach for FSI analyses of adam-reservoir-computation time, while allowing for more advanced features such as bottom absorption and non- be a challenging task, where it has a significant impact on the results. The method proposed by Spears and Jensen has a physical meaning in the sense that this method excites the same e ective mass for the Rayleigh damped case as for the modal damped case. If a constant modal damping is desired or prescribed in a standard, this method provides a reasonable and sound method to choose the Rayleigh damping parameters for a complex structure. A more straightforward method is to choose the two frequencies in such a way that the span between the frequencies covers about 80% of the e ective mass. The choice of reservoir boundary conditionsparameter showed to be the one that least a ected the results in the time-history analysis. results and this coe cient should be used carefully.

  • Smirnov, Serguei
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS), Micro and Nanosystems.
    Lioubtchenko, Dmitri
    KTH, School of Electrical Engineering and Computer Science (EECS), Micro and Nanosystems.
    Oberhammer, Joachim
    KTH, School of Electrical Engineering and Computer Science (EECS), Micro and Nanosystems.
    Single-walled carbon nanotube layers for millimeter-wave beam steering2019In: Nanoscale, ISSN 2040-3364, E-ISSN 2040-3372Article in journal (Refereed)
    Abstract [en]

    The ability to efficiently transmit and manipulate high-frequency signals poses major challenges resulting in a lack of active and reconfigurable millimeter-wave and terahertz devices that are needed to enable beyond-5G broadband communication systems. Here, thin single-walled carbon nanotube (SWCNT) layers are introduced as a tunable impedance surface for millimeter-waves. Carbon nanotube layers are integrated with dielectric rod waveguides. Their surface impedance, tuned by light, is shown to modify the wave propagation inside the waveguide. A direct application of the effect is a phase shifter, demonstrated experimentally and by numerical simulations. Additionally, an antenna array of two dielectric waveguides, one covered in SWCNTs, is designed and fabricated. The proof-of-concept illustrates optically-controlled beam steering enabled by carbon nanotubes, and directions for further device optimizations are provided. These findings demonstrate thin SWCNT layers as an optically-reconfigurable element, suitable for broadband millimeter-wave communications.

  • Jin, Jacky
    et al.
    KTH, School of Architecture and the Built Environment (ABE), Civil and Architectural Engineering, Building Technology and Design.
    Matskin, Artur
    KTH, School of Architecture and the Built Environment (ABE), Civil and Architectural Engineering, Building Technology and Design.
    Hållbara ytbeläggningar i parkeringshus: En undersökning av olika beläggningssystem efter 5 till 10 års drift2019Independent thesis Basic level (university diploma), 10 credits / 15 HE creditsStudent thesis
    Abstract [en]

    Parking facilities are constantly exposed to traffic loads from driving vehicles and chloride attacks during the winter period. The consequence of this is costly damage to the concrete structure, especially in the event of cracks. Chloride penetration causes the reinforcement bars to corrode and weaken the underlying concrete structure. Replacement of the corroded reinforcement bars and concrete repairs are costly for Stockholm Parking. The solution is to pretreat the unprotected concrete and apply a coating above it. The most common coatings found in Sweden are cement-based, bitumen-based and thermoset-based coatings. There is no clear choice when it comes to coatings since they all have different properties. The advantages and disadvantages of different coatings that the customer must value depend on such conditions of the parking garage as the shaping, the traffic load and whether it is the base plate or the intermediate floor. Besides the material, the execution is also important. A well done pretreatment of the concrete substrate and meticulous laying of the coating reduces the possibility of the occurrence of damage in the future and thereby the repair costs. The purpose of the thesis project was investigation of various coatings systems that have been in operation for 5 to 10 years and identification and mapping damages that occur to them. In this report, five parking facilities have been selected as case studies to investigate the most common coatings used today in Stockholm Parkering parking facilities. Based on literature studies, interviews with various parties in each project, observations made on study visits and comparison between case studies conclusions were drawn about the most sustainable surface coating systems. Finally, a guidance is created to simplify the choice of coating systems for parking facilities with different conditions.

  • Sjölund, Magdalena
    et al.
    KTH, School of Architecture and the Built Environment (ABE), Civil and Architectural Engineering, Building Materials.
    Stenis, Nelly
    KTH, School of Architecture and the Built Environment (ABE), Civil and Architectural Engineering, Building Materials.
    Limträhybrider av acetyleradbjörk och obehandlad gran: En experimentell och numerisk analys av vissamekaniska egenskaper och fuktbeteende2019Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [sv]

    I Sverige finns en lång tradition kring användandet av trä för konstruktionsändamål. I den storskaligabyggindustrin har det dock kommit att handla mycket om stål och betong och detta gäller inteminst för konstruktioner utomhus såsom broar. I träkonstruktioner har speciellt infästningarna storbetydelse då dessa är svåra att skydda från fukt, vilket gör att träets begränsande beständighetblir en anledning att välja andra material. Genom kemisk modifiering av trä, så kallad acetylering,kan detta beständighetsproblem lösas, dock med risk för kraftigt ökade materialkostnader. Syftetmed detta arbete var att undersöka möjligheterna kring så kallade limträhybrider av kemiskt modifieradbjörk och omodifierad gran. Tanken är att erhålla god beständighet i svårskyddade delarav en limträbalk, t.ex. vid knytpunkter, utan att behöva använda modifierat trä i hela balken.Genom att fingerskarva samman de två materialen kan det acetylerade materialet användas justintill infästningen och sedan övergå till obehandlat trä i områden som lättare kan skyddas från riskför röta med hjälp av till exempel konstruktionstekniskt träskydd. Mer specifikt handlar arbetetom att studera hur en sådan typ av hybridkonstruktion beter sig när den testas i fyrpunktsböjning.Speciellt undersöktes hur och var brottet uppkom samt balkarnas böjhållfasthet och styvhet.Dessutom undersöktes, med hjälp av småskaliga tester, hur acetyleringen av björk påverkar dessE-modul och hållfasthet.Studien är baserad på både experimentella försök och numerisk analys. Fyrpunktsböjning användesvid försök på hybridbalkarna och materialtesterna. Finita elementmetoder användes för attförutspå var brott kommer att uppstå samt hur området vid fingerskarvarna påverkas av den storaskillnaden i svällningsegenskaper mellan obehandlat trä och det mer dimensionsstabila modifieradeträet.Resultaten indikerade att den acetylerade björken hade större spridning med avseende på det undersöktamaterialegenskaperna jämfört med omodifierad björk. Testerna av hybridbalkarna visadepå stora problem kring fingerskarvarna beroende på problem vid limningen och även lokala defekter.Resultaten visade även en klar förbättring gällande balkarnas styvhet då limträbalkarnasmittpartier bestod av björk.Även de finita elementberäkningarna indikerade att brott kan förväntas i fingerskarvarna samtatt problem kan uppstå som följd av att limträet av omodifierad gran sväller vid uppfuktning ijämförelse med limträet av acetylerad björk. Vidare forskning föreslås bland annat kring utvecklingav lim som lämpar sig för såväl barr- som lövträ, obehandlat- som acetylerat material samtfingerskarvning. Resultaten av studien visar på goda utsikter att förbättra limträets styvhet ochböjhållfasthet samt en möjlighet till att lösa problem med fukt kring infästningar.