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.

Senast inlagda

Bild (thumbnail)
Post
A Framework for Minimising Java Microbenchmark Suites - A Metric-Based Approach
(2024) Darin Nordqvist, Filip; Otterlind, Rasmus; Chalmers tekniska högskola / Institutionen för data och informationsteknik; Chalmers University of Technology / Department of Computer Science and Engineering; Gay, Gregory; Leitner, Philipp
Performance is an important non-functional requirement. The success of software projects is highly dependent on the adequate performance of the software. However, one fundamental problem with performance testing is that it can become immensely time-consuming to conduct. In this paper, several different code metrics were explored and used for conducting test case minimization (TCM) of Java Microbenchmark Harness (JMH) suites. The microbenchmark suites were created with a tool called ju2jmh, which creates microbenchmark suites based on already existing unit testing suites of a project. The metrics used for conducting TCM were lines of code (LOC), loop count, cyclomatic complexity, and the combination of these metrics. The results indicate that suites derived using LOC and loop count performed worse than randomly sampled suites, while suites derived from the cyclomatic metric and the combination of LOC, loop count, and cyclomatic complexity performed similar or slightly worse than randomly sampled suites. Furthermore, it is not infeasible to assume that if these metrics were refined by utilising dynamic techniques, some of them could potentially outperform randomly sampled suites consistently.
Bild (thumbnail)
Post
Requirement representation for safety critical and fairness aware automotive perception systems - Identifying requirements representation challenges for multi party collaboration.
(2024) Jakobsson, Oskar; Rohacova, Zuzana; Chalmers tekniska högskola / Institutionen för data och informationsteknik; Chalmers University of Technology / Department of Computer Science and Engineering; Penzenstadler, Birgit; Heyn, Hans-Martin; Saeeda, Hina
Background: Advancements in the field of machine learning (ML), have unlocked new capabilities for Driving Automation Systems (DAS). These DAS systems rely on the input from automotive perception systems. These systems exist in a safety critical domain, and well-defined requirements are key to ensure that they can operate safely. However, requirements engineering (RE) for ML-enabled systems has been identified as a challenge in research. Aim: The thesis aimed to investigate current approaches in RE for automotive perception systems, and identify what challenges exist and which processes work well. Specifically, approaches for requirement representations, model kinds, templates, and structures. The thesis also wanted to explore if a shared language, in regards to domain description, reference system architecture, and reference information model, could help mitigate potential challenges without hindering approaches that work well. Methods: An exploratory case study was conducted by interviewing experts in the automotive field. This included participants from a major automotive OEM, suppliers to said OEM and experienced researchers in the automotive field. In total ten interviewees were consulted. Results: The challenges and what works well in the current processes in the case study companies were identified through thematic analysis of the interview data. The thesis explored the potential of a shared language to mitigate these challenges by discussing the topic with interviewees and observing brainstorming workshops for the creation of the shared language. Conclusions: The thesis shows that there is a lack of industry standards in RE for ML-enabled automotive perception systems, which complicates multi-party development. According to interviewees, the shared language has potential to alleviate the identified challenges. However, the feasibility of the shared language is still unclear.
Bild (thumbnail)
Post
Multi-agent Communication via Reinforcement Learning in Social Networks
(2024) Liang, Zhitao; Wang, Wanqiu; Chalmers tekniska högskola / Institutionen för data och informationsteknik; Chalmers University of Technology / Department of Computer Science and Engineering; Dubhashi, Devdatt; David Thomas, Jonathan; Carlsson, Emil
This thesis investigates the use of multi-agent reinforcement learning (MARL) to explore emergent communications of artificial agents in social networks. The main goal is understanding how agents develop shared communication protocols to perform collaborative tasks in complex environments. Using the World Color Survey (WCS) dataset, we implement a speaker-listener model in which an agent learns to name colors, providing a framework for observing the formation of communication strategies. In contrast to existing work, we utilize a shared neural network for both speaker’s and listener’s functions, which promotes equivalence in language use between agents and supports consistent communication. Extending the model to multiple agents, we studied how social network structure affects emergency communication, finding that denser networks produce more consistent language while sparser networks allow for greater diversity. The introduction of new agents and different levels of interaction between communities also affects language evolution, with newly generated languages found to be more similar to more populous collectives. However, the scale of our research could be improved. In future work, investigating larger populations of agents would be beneficial for better understanding scalability and refining our findings. Additionally, we could explore other communication modes, such as one to-many or many-to-one interactions, to gain a more comprehensive understanding of emergent communication in artificial systems.
Bild (thumbnail)
Post
Studying an Architectural Pattern for Deep Learning training code - Collecting and Addressing Current Software Quality Issues within Academia and the Automotive Industry for Deep Learning
(2024) Razaq, Behroz; Johansson, Sebastian; Chalmers tekniska högskola / Institutionen för data och informationsteknik; Chalmers University of Technology / Department of Computer Science and Engineering; Penzenstadler, Birgit; Heyn, Hans-Martin
Deep learning has become more popular throughout the years, consequently, an expansion of new developments within the field has occurred. As deep learning is mainly practiced by writing code, many established software engineering practices can be transferred to the field. While this has happened to some extent in some areas, like requirement engineering and MLOps, other subfields have lagged behind. Writing reusable and modular code is important for easy development, but there does not seem to exist a convention for how to write such code for deep learning training. Therefore, the architectural pattern MODLR was created and in this thesis, it was analyzed against found problems from practitioners of deep learning. One of the main goals of MODLR is to decouple loss code and to show the relevance of this focus, GitHub repositories were mined and automatically categorized projects based on their loss code, with the help of an LLM. The results show that MODLR is a good fit for an architectural pattern within the space of deep learning. As a bonus, it also shows one of the ways LLMs can be used to help research with automatic large-scale analysis of code.
Bild (thumbnail)
Post
Understanding Customer Value - Identify Customer Value in Existing Features, Products and Systems to Inform Decision-Making
(2024) Holmesten, Felix; Thim, Lukas; Chalmers tekniska högskola / Institutionen för data och informationsteknik; Chalmers University of Technology / Department of Computer Science and Engineering; Penzenstadler, Birgit; Bosch, Jan
This thesis presents the development and validation of a new framework aimed at addressing the challenges in the field of customer value and product management. Through a comprehensive literature review and interviews with industry professionals multiple key challenges were identified that formed the basis of for the framework. The framework itself focuses on understanding what constitutes value in existing features, products, or systems utilizing hypotheses and metrics to conduct experiments. The experiment results are then used to predict and calculate the expected value of new features. Conducting a validation workshop demonstrated the effectiveness of the framework, guiding the participants to a better understanding of value and successfully mitigating some of the identified challenges. Despite the successes of the framework, it also acknowledges certain limitations and presents opportunities for refinement and future research. Nevertheless, the frameworks accessibility and potential for practical applicability in industry highlights the contributions it brings to the field. The framework presents a practical solution to challenges in the field of customer value and product management, with its potential benefits recognized and validated by industry professionals.