No description available.

Impressive results can be achieved when stacking deep neural networks hierarchies together. Several machine learning papers claim state-of-the-art results when evaluating their models with different accuracy metrics. However, these models come at a cost, which is rarely taken into consideration. This thesis aims to shed light on the resource consumption of machine learning algorithms, and therefore, five efficiency metrics are proposed. These should be used for evaluating machine learning models, taking accuracy, model size, and time and energy consumption for both training and inference into account. These metrics are intended to allow for a fairer evaluation of machine learning models, not only looking at accuracy. This thesis presents an example of how these metrics can be used by applying them to both text and image classification tasks using the algorithms SVM, MLP, and CNN.

Heart arrhytmias are rhythm disorders, which affects the heart and can lead to stroke, hospitalization, and a lower quality of life. Arrhythmias are often detected and diagnosed by using an electrocardiogram (ECG) and in many cases, the early detection of an arrhythmia can greatly increase the chance of a better recovery. The goal of this project is to develop a prototype of a portable ECG-system which can record and classify ECG-signals in real-time so that arrhythmias can be detected as early as possible. The prototype is made up of 1) a wireless ECG measurement unit consisting of a micro-controller, electrodes, an analog front-end, and a Bluetooth module, to capture ECG-signals from a person and transmit them via Bluetooth to a smartphone; and 2) an application implemented on a smartphone that is used to record/receive, visualize and classify ECG-signals. The classification is performed using a machine learning technique - a modular convolutional neural network, which is developed and trained by using a data set of ECG-signal features that is based on R-peaks. The prototype has been developed and tested. The test results have shown that the prototype is fully functional in terms of ECG-signal acquisition, transmission and classification. The results of evaluating the classifier show a weighted average recall of 98%, and a weighted average precision of 98%. The smart phone application used to record and classify the ECG-signals is shown to have a small footprint on the energy consumption of the smartphone, allowing for lengthy recordings. The performance of the classifier can be further improved through the use of a larger and more balanced data set.

The main goal of this project is to evaluate and demonstrate the feasibility of a new framework to be used to design and control external simulators at SAAB AB. The project explore the possibility of developing a web-based framework in a secure environment, based on Docker technology and modern application frameworks. In order to reach the main goal, a qualitative interview is conducted, which is used to design the new framework. An evaluation of the feasibility of the new framework is performed. For that a proof of concept has been developed, together with a measure of performance and the return of investment of the new framework. With the proof of concept together with the performance evaluation, it is proved that a new web-based framework is suitable to design and control external simulators. Furthermore, the return of investment is estimated to be 76.73% per year, which suggests that the benefits would succeed the investment cost of the new framework, if used for more than 16 months.

With the increasing trend of using cloud computing resources, the efficient utilization of these resources becomes more and more important. Working with data stream processing is a paradigm gaining in popularity, with tools such as Apache Spark Streaming or Kafka widely available, and companies are shifting towards real-time monitoring of data such as sensor networks, financial data or anomaly detection. However, it is difficult for users to efficiently make use of cloud computing resources and studies show that a lot of energy and compute hardware is wasted. We propose an approach to optimizing resource usage in cloud computing environments designed for data stream processing frameworks, based on bin packing algorithms. Test results show that the resource usage is substantially improved as a result, with future improvements suggested to further increase this. The solution was implemented as an extension of the HarmonicIO data stream processing framework and evaluated through simulated workloads.

Implementing an indoor navigation system requires alternative techniques to the GPS. One solution is vision-based localization which takes advantage of visual landmarks and a camera to read the environment and determine positioning. Three computer vision algorithms used for pose estimation are tested and evaluated in this project in order to determine their viability in an indoor navigation system. Two algorithms, SIFT (Scale-Invariant Feature Transform) and SURF (Speeded-Up Robust Features), take advantage of the natural features in an image, whereas the third algorithm, ArUco, uses a manufactured marker. The evaluation displayed certain advantages for all solutions, however with the goal of using it for a navigation system ArUco was the superior solution as it performed well for key criteria, mainly computational performance and range of detection. An indoor navigation system for Android devices was developed using ArUco marker tracking for positioning and augmented reality for pro- jecting the route. The application was able to successfully fulfill its goal of providing route guidance to a specific target location.

På grund av begärd sekretess innehåller följande rapport enbart projektets bakomliggande problemformulering samt de slutsatser som framtogs.

No description available.

Klimatfrågan blir allt viktigare i dagens samhälle. Halten av koldioxid har ökat i atmosfären och är nu det högsta på 800 000 år. Ökningen beror på den mänskliga faktorn samt den ökade förbrukningen av fossila bränslen de senaste 70 åren. Med detta i beaktande har uppdragsgivaren till examensarbetet efterfrågat hur användning av historisk data kan ligga till grund för miljömässiga beslut.   Den här fallstudien avser att beräkna koldioxidutsläpp från lastbilstransporter med hjälp av beräkningsverktyget NTM. Koldioxidutsläpp har beräknats mellan distributionslager i Jönköping till terminal i Sundsvall samt leveranser från terminal ut till kund, the last mile.  Studien bygger på en empirisk fallstudie där kunskapen grundas på verkliga observationer från historisk data som ligger till grund för beräkningarna. Informationsinsamling sker genom mixed-method, vilken kombinerar kvalitativa och kvantitativa studier.   I resultatet beräknas koldioxidutsläpp på direkttransporter och mjölkrundor. Resultatet påvisar att mjölkrundor är mer effektivt jämfört med direkttransporter med avseende på miljön. Vidare har ett samband tagits fram för att ta reda på vad respektive sändnings har för koldioxidbidrag i den avsedda rutten. Detta genereras ett approximativt värde som kan tillämpas av uppdragsgivaren. / Climate issues are becoming increasingly important in today's society. The carbon dioxide content has increased in the atmosphere and is now the highest in 800 000 years. The increase is due to the human factor and the escalation of fossil fuel consumption in the last 70 years. With this in mind this bachelor’s thesis has been requested by the employer to use historical data which can be the reason for environmental decision taking in the future.   This case study aims to calculate carbon dioxide emissions from truck transport using the NTMcalc. Carbon dioxide emissions have been calculated from the distribution warehouse in Jönköping to the terminal in Sundsvall and deliveries from the terminal to the customer, in other words the last mile. The study is based on an empirical case study where knowledge is based on real observations from historical data as the foundation for the calculations. For research of information the method mixed-methods have been applied. This method combines qualitative and quantitative studies.   In the result, emissions are calculated on direct transport and milk runs. The result demonstrates that milk runs are more efficient compared to direct transport regarding to the environment. Furthermore, a correlation has been developed to find out the contribution for each transmission in the intended route. This is generated by an approximate value that can be applied by the employer in the future.

No description available.