Chalmers Open Digital Repository

Välkommen till Chalmers öppna digitala arkiv!

Här hittar du:

Studentarbeten utgivna på lärosätet, såväl kandidatarbeten som examensarbeten på grund- och masternivå
Digitala specialsamlingar, som t ex Chalmers modellkammare
Utvalda projektrapporter

Forskningspublikationer, rapporter och avhandlingar hittar du i research.chalmers.se

Enheter i Chalmers ODR

Välj en enhet för att se alla samlingar.

Visar 1 - 2 av 2

Senast inlagda

LLM-based Log Analysis for Fault Localization in the Automotive Industry

(2025) Ekström, Anton; Rhedin Stam, Hampus

This thesis investigates the application of large language models (LLMs) to aid practitioners of log analysis for fault localization in the automotive industry. An existing LLM-based log summarization tool is extended and evaluated, focusing on the cognitive load of practitioners and how satisfied they are with the tool. The effect of LLM-based log summarization on productivity of practitioners in the automotive industry is investigated through a case study at a company within the automotive industry. Think-aloud sessions and semi-structured interviews are carried out to asses the impact of the tool on the fault localization flow of study participants. Results suggest that LLM-generated log summaries can aid practitioners by giving them a first glance of the issue, thereby potentially reducing manual effort and improving productivity. However, the results also suggest that the context of the issue, domain knowledge, and interactivity of the tool plays a major role for success. A lack of context and means for the practitioner to guide the tool could result in a less effective workflow with higher cognitive load. The thesis provides insights on the integration of LLM-based log analysis tools within fault localization workflows in the industry, highlighting both the benefits and challenges of deploying LLMs in real-world fault analysis scenarios

Emerging Architectures for Chemical Language Modeling

(2025) Hagström, Ester; Redmo Axelsson, Erik

In recent years, language modeling architectures have become increasingly prominent in the field of generative chemistry, offering new approaches for the de novo design and optimization of small molecules. This thesis presents a comparative study of two emerging architectures: the decoder-only Transformer and the Mamba architecture, and a conventional Recurrent Neural Network with LSTM cells. The investigation explores how choices in training data, including a targeted medicinal dataset (ChEMBL) and a chemically broad dataset (PubChem), as well as data augmentation via randomized SMILES representations, influence generative capacity and chemical space coverage. In addition to this, task-specific optimization of models through reinforcement learning is studied, and the models are compared with respect to their ability to generate diverse molecules with desired properties. Through pretraining experiments, it is shown that while the Mamba and RNN architectures reach their optimum performance significantly faster, the decoder-only Transformer achieves the highest validity and uniqueness in molecular generation. Training on PubChem, as opposed to ChEMBL, generally enhances validity and uniqueness but tends to reduce novelty, indicating a trade-off between chemical space saturation and innovation. As for data augmentation through randomization of SMILES, this helped all models refrain from memorizing the dataset, resulting in higher novelty across architectures and datasets. Reinforcement learning experiments further reveal that all three architectures are capable of optimizing toward specific molecular properties, with the decoder-only Transformer and Mamba each exhibiting distinct strengths depending on the optimization task. Regarding the pretraining condition’s effect on reinforcement learning, ChEMBL-trained models outperformed those trained with PubChem on multiple tasks, and all architectures, but especially Mamba, benefitted from being pretrained with randomized SMILES. Notably, even reduced-parameter models, such as a downsized decoder-only Transformer variant, perform competitively relative to larger architectures.

Applied ratio pyrometry for temperature measurements in industrial-scale flames

(2026) Raffai, Herman

Optimizing Stream Engines for use in eFPGAs on Radiation Hardened SoCs

(2025) Magnusson, Adam; Örtenberg, Erik

Systems on a Chip (SoCs) are becoming increasingly common for use in most computational domains as heterogeneous hardware architectures prove themselves very efficient and powerful. The space domain is one such example and poses a plethora of challenging design constraints, which become even more pronounced in the context of radiation hardened embedded Field Programmable Gate Arrays (eFPGAs). eFPGAs lend themselves to supporting powerful hardware accelerators (HAs), where data is streamed in by the use of a stream engine. Due to the relatively small amount of programmable logic in eFPGAs, the stream engine supporting the HA must be made as resource efficient as possible. Though, to the best of the authors knowledge, there is no previous work exploring resource optimized stream engines for use on eFPGA. In this thesis, we implement a performance and resource efficient stream engine for eFPGAs. The proposed stream engine – named GANIMEDE – achieved a communication link utilization of 94.5% accounted for protocol overheads while only occupying 2.1% of the total available resources provided by the targeted eFPGA. In addition, this thesis offers a discussion on desirable properties in stream engines and presents parts of supported protocols that should be implemented on the SoC instead of the eFPGA.

Effects of Cognitive Load in Human-AI Requirements Engineering

(2025) Shivamurthy Praveen, Niharika Nandi; Sasvihalli, Laxmi Prashantraddi

As Artificial Intelligence becomes more integrated into software engineering, its role in decision-support systems within Requirements Engineering has grown. However, the cognitive demands placed on users interacting with these AI tools remain underexplored. This thesis investigates how explanation formats offered by Explainable AI affect mental effort, task difficulty, confidence, and correctness during requirements engineering inspired prioritization tasks. Through a controlled experiment with 61 participants, three XAI formats of bar charts, textual explanations, and confidence scores were evaluated across two task pairs of differing complexity. The study examined the influence of task complexity and explanation format, the impact of explanation type on decision-making quality, and whether participant preferences for certain formats aligned with improved performance and lower cognitive strain. Statistical analyses, including Spearman correlation and independent t-tests, revealed that task complexity consistently influenced cognitive load, while explanation format had no clear effect. Additionally, although preferred formats did not universally enhance task performance, participants who favored confidence scores showed marginally higher correctness and confidence levels. These findings suggest that cognitive effort in AI-assisted requirements engineering tasks is shaped more by task characteristics than explanation format alone, and that tailoring explanations to individual user preferences may offer subtle benefits.