Examensarbeten för masterexamen
Länka till denna samling:
Browse
Senast publicerade
- PostInfluence of Track-Assembly Geometryon Mobility Over Rigid Terrai(2024) Körner, Alexander; Chalmers tekniska högskola / Institutionen för matematiska vetenskaper; Andersson, Adam; Karlsson Hagnell, MathildaThis thesis considers the modeling and simulation of tracked vehicles. Specifically it aims to investigate how the geometry of the track-assembly affects its mobility over rigid obstacles. This is done with the purpose of aiding in the Swedish Defence Research Agency’s (FOI) goal of developing parametrized virtual models of tracked vehicles for researching mobility. The vehicle is modeled as a multibody dynamics simulation consisting of rigid bodies. This entails solving index-3 differential algebraic equations (DAE) with high precision. The validation of the vehicle model and simulation included the comparison of two methods of solving the DAE in question, both of which require the solving of a saddle point problem. Thus for each method two iterative saddle point problem solvers have been compared as well. After choosing a numerical method the mobility of the model is evaluated on rigid obstacles with varying track-assembly geometry. The evaluation shows that the track-assembly geometry significantly impacts the models ability to traverse vertical step obstacles and half round obstacles. The results suggest that this applies to a general real tracked vehicle, however a more thorough model validation and investigation into the numerical methods is required to draw any quantitative conclusions.
- PostPharmaceutical assay search with AI(2024) Alladin, Ali; Chalmers tekniska högskola / Institutionen för matematiska vetenskaper; Jonasson, Johan; Jonasson, JohanRetrieving historical assay data in pharmaceutical research is often restricted by reliance on specific metadata, overlooking the contextual information in associated protocol documents. This thesis investigates the potential of utilizing these plain English protocol documents alongside Natural Language Processing (NLP) techniques to implement semantic search for assays. A baseline TF-IDF model and the Transformer models BERT, SBERT, and Longformer were used to get embeddings of protocol documents from a corpus of historical protocols. Their performance in retrieving relevant historical protocols was evaluated based on key technical criteria, where the TF-IDF models and BERT using the chunking technique showed the best results. However, limitations in the evaluation scope introduce some uncertainty to the findings, highlighting the need for more rigorous validation. Nevertheless, the conclusions suggest that integrating NLP-driven semantic search systems could reduce the time and manual effort required for assay retrieval, even though the current approach may need further refinement for practical application. These insights are a promising foundation for developing AI-powered search systems used for pharmaceutical texts.
- PostLearning-Based Detection of Events in Eye-Tracking Data: An Investigation Into Small-Scale Models for Automotive Applications(2024) Due, Martin; Chalmers tekniska högskola / Institutionen för matematiska vetenskaper; Ringh, Axel; Molin, Vincent; Zemblys, RaimondasDetection of eye movement events in eye-tracking data is integral to various research fields and commercial applications. Traditionally, the detection has been accomplished either by hand or with threshold based detectors with the inherent drawback that thresholds levels and other parameters had to be hand picked for each scenario. In recent years machine learning methods have been employed for eye movement event detection that do away with that requirement. However, these models tend to be too large to run on limited resources in embedded applications, particularly automotive ones. This thesis focuses on creating models that are smaller but with retained performance. Five machine learning methods were evaluated and hyperparameters were tuned to create well performing small models. Furthermore, the usage of synthetic data for training was investigated, both as a supplement to real data and as a sole source of training data. The study found that a Multilayer Perceptron model (MLP) trained on a combination of real and synthetic data struck the best balance between size and performance. Additionally, results show that models trained purely on synthetic data also performed reasonably well. The findings of this thesis suggest that small, efficient models can effectively detect eye movement events, with potential applications in automotive contexts.
- PostDO SWEDISH MUTUAL FUNDS DESERVE THEIR FEES?- A statistical resampling approach(2024) Brantberg, Jesper; Chalmers tekniska högskola / Institutionen för matematiska vetenskaper; Mostad, Petter
- PostPatient outcome prediction using knowledge graph representation learning(2021) Fazlinovic, Adnan; Modi, Trilokinath; Chalmers tekniska högskola / Institutionen för matematiska vetenskaper; Lundh, Torbjörn; Kjellberg, Magnus; Lundh, TorbjörnThe project focuses on using knowledge graphs in a healthcare setting, classifying patient re-admissions. Knowledge graphs are a type of heterogeneous network consisting of entities and relations. Knowledge graph embedding method aims to generate lower-dimensional latent vector representation of these entities and relations while preserving their relational properties. The data consists of patient admission details along with their underlying diagnoses, prescriptions consumed and procedures performed. To exploit the true nature of knowledge graphs, more information to the patient graph is added by combining various biomedical databases to obtain a richer set of relationships. Cleaning patient records and converting the information in more standardized form, as well as gathering information and create a knowledge graph structure in the form of triplets are conducted. The generation of latent vector representations of the entities and relations are done with various embedding methods, where the final phase is to classify patient re-admissions. The methods investigated achieves to represent entities and relations in latent vector form when evaluating the embeddings based on the proposed loss functions. However, the embeddings generated doesn’t supply enough information that can accurately predict the patient readmission status in an extended down-stream fashion. The potential problems could be either of not enough features that explains the variability, not enough rich information regarding the different data sources used, or the effect of class imbalance. A stratified test subset was created from the same excerpt of training data to quantify the results.