Chalmers Open Digital Repository
Not a member yet
26247 research outputs found
Sort by
Symmetry-Embedded Siamese Neural Networks for Regression Tasks
No full textMachine learning (ML) has gained momentum in early drug discovery by reducing the number of expensive and time consuming real-world experiments. One class of ML that has recently shown promise are siamese neural networks (SNN) which take a pair of inputs and predicts the difference in a property, rather than the absolute property. It follows that such a model should be anti-symmetric with regards to the order of the inputs. We introduce a new design for an SNN that has this symmetry embedded directly into the architecture to increase the reliability of the model. By pairing a test compound with a series of reference compounds the model can produce an ensemble of predictions where the mean can be treated as the final prediction and the variance can be used as an uncertainty measure. The symmetry embedded SNN (SE-SNN) shows comparable performance to baseline models on five chemical datasets
Age-Dependent Material Modelling of the Human Cranium: Predicting Cortical Bone Response Using Bayesian Networks and Experimental Validation
No full textThe risk of skull fracture due to impact increases with age. This project investigates how age affects
the mechanical behaviour of cortical bone in the human skull, with the aim of enhancing the accuracy
of finite element head models used in fracture prediction. Material properties were derived from
experimental data by Wood (1969), which included tensile tests on post-mortem human subjects
(PMHS) across a wide range of ages. A Bayesian statistical model was developed in order to analyse
the influence of age on strain rate, and material properties such as modulus of elasticity, breaking
stress, and breaking strain. The results showed a strong correlation between strain rate and modulus
of elasticity, while direct correlations with age were inconclusive due to wide posterior distributions.
Stress-strain curves were generated based on the Bayesian model and used to create age-specific material
cards in LS-DYNA simulations. These curves accounted for varying strain rates, allowing more
realistic simulation of impacts. To validate the finite element models, simulations were compared
to experimental impact data from Raymond et al.(2009), in which PMHS were subjected to blunt
impacts to the side region of the head. Fracture risk was evaluated using a survival analysis-based
approach for creating risk functions, and compared against experimental outcomes. Although some
prediction discrepancies were observed, the trends suggest that aged individuals experience slightly
higher strain under the same loading conditions, implying an increased risk of fracture.
The study demonstrates the potential for age-sensitive material models to improve simulations of
the human head. It also highlights the need for updated experimental datasets and additional
consideration of other anatomical variables such as skull and scalp thickness in future wor
Opportunities and Feasibility of Urban Freight Consolidation: A Case Study in the Event Area of Gothenburg
No full textWith economic development and the growth of tourist-based event areas, urban freight transport faces significant challenges in terms of congestion and environ-mental impact. The freight activities in the event area of Gothenburg, especially Liseberg and Svenska Mässan, are characterized by fragmented shipments, seasonal fluctuations and spatial constraints, leading to logistical inefficiencies. This study combines stakeholders’ interviews, freight trip generation (FTG) analysis and cost-based modeling to examine current delivery patterns in the event area, analyze the potential for freight consolidation, and assess the feasibility and sustainability of implementing consolidation strategies with an urban consolidation center (UCC). The results show that there is a strong need for freight consolidation in the area due to fluctuating, small-volume, frequent, and low-loading-rate deliveries, and that approaches such as multi-stop deliveries and receiver-led consolidation can signifi-cantly reduce delivery costs and traffic flows. Finally, the study provides actionable recommendations for stakeholders, including suppliers, logistics service providers, receivers, and public authorities, to advance consolidation strategies. This research contribute to the REDIG project’s goal of developing scalable, fossil-free logistics solutions that improve urban transport efficiency and support sustainable urban freight in Gothenburg and beyond
Datadriven kvantfelskorrigering genom avkodning av repetitionskoden med grafneurala nätverk
No full textQuantum error correction is one challenge faced when implementing quantum computer
systems. Repetition code is a simpler correcting decoder that detects bit- or phase-flip errors.
To fully benefit from the repetition code, an appropriate decoder is required. Therefore,
this report aims to compare two different decoding methods, a classical algorithm known as
Minimum-Weight Perfect Matching, MWPM and a machine learning-based method using
Graph Neural Networks, GNN. A modular phase-flip detecting repetition code is constructed
and executed on an IBM Quantum processor. The measured syndromes are used as training
data for GNN, which is trained on Chalmers high-performance computing cluster Alvis.
The report also investigates how decoding time scales with increasing code area. Results
showed that the GNN decoder achieved marginally higher logical accuracy αL compared
to MWPM, particularly near the code area where optimization of the hyperparameters
was performed. Furthermore, GNN achieved slightly higher logical accuracy compared
to MWPM for code distance and time repetitions d,dt ≤ 9. This is because the decoder
was tested on syndromes generated during the same quantum computer execution as the
training data. However, GNN also showed consistently higher logical error probability pL
and was less effective than MWPM when tested on different syndromes, which may hinder
practical application in environments with varying error distributions
Impedance measurement techniques for DC-biased systems
No full textAbstract
This thesis compiles information about six methods for impedance measurement of DC biased systems. It describes a physical interface for low-voltage Devices Under Test (DUTs) using the shunt-through method with DC-blocking capacitors. The interface is used to measure impedance versus frequency of three pouch cells of different dimensions and chemistries. In addition to the cell measurements, a separate set of measurements are carried out on inductors and capacitors with and without the use of the DC-blocking interface. The measurements show how parameters such as heat, conductor length and measurement setup affect the measured impedance. An interface model is presented that explains in what frequency and DUT impedance ranges each parasitic interface parameter affects the measurement results, and in what way. The conclusion of the thesis is that the used capacitor-based impedance measurement method was not accurate when measuring the low impedance of an electric vehicle battery cell. However, with the modification and improvements outlined in the thesis, the interface type could be used to measure devices with larger impedance, such as a complete high voltage electric vehicle battery consisting of the same cells
Spider-Scents v2: Enhancing Gray-Box Scanning for Stored XSS Vulnerability Discovery
No full textStored XSS vulnerabilities pose significant security risks in modern web applications, yet detecting them remains challenging due to their complex data propagation paths and the limitations of traditional scanning tools. Spider-Scents addresses these challenges using a gray-box database-aware approach that directly injects payloads into backend storage, effectively bypassing the difficulties of conventional input-based fuzzing. Building on this foundation, we present Spider-Scents v2 — a substantial enhancement of the original Spider-Scents prototype. Spider-Scents v2 introduces two key advancements: a refined table traversal strategy that models the database schema as a directed graph and leverages BFS and DFS to systematically explore injection paths; and a format-aware payload customization module specifically tailored for JSON-structured data, which is increasingly common in modern applications. To evaluate the practical impact of these enhancements, we conduct a series of experiments assessing Spider-Scents v2’s performance on both the original PHP-based applications and a broader set of non-PHP applications. Furthermore, we investigate how Spider-Scents database synthesis algorithm can serve as a preparatory module to augment the vulnerability detection capabilities of other black-box scanners, including Black Widow, Burp Suite, ZAP, and SCNR. Our results demonstrate that Spider-Scents v2 offers measurable improvements in stored XSS detection, achieving better coverage and uncovering new vulnerabilities that were previously undetected. Additionally, the integration of Spider-Scents data synthesis algorithm enhances the effectiveness of third-party scanners, highlighting its potential as a complementary tool for web security assessments
Patent Country Selection Strategy for Global Automotive OEMs
No full textIn knowledge-based economies, patents play a central role in protecting innovation and enabling strategic positioning. In the automotive sector, characterized by complex products, concentrated innovation clusters, and profound technological transformation, deciding where to file patents is a growing strategic challenge. This thesis explores how a global automotive original equipment manufacturer (OEM) can approach patent country selection in a way that supports the firm’s broader intellectual property (IP) strategy and responds to changes in the competitive and technological landscape.
Using a qualitative case study method, data was collected through sixteen semi-structured interviews with internal stakeholders and external IP professionals, complemented by professional literature and practitioner insights. The analysis was guided by Somaya’s (2012) framework of generic patent strategies, proprietary, defensive, and leveraging, along with theories on technological change, innovation ecosystems, and patent strategy.
The findings show that market relevance and third-party exposure are the most critical factors in country selection, while legal system quality is treated as a baseline requirement. The strategic intent of the patent portfolio significantly shapes which factors are prioritized. For proprietary strategies, filings focus on key markets; defensive strategies target both own and competitors’ key regions; and leveraging strategies emphasize collaborators’ and suppliers’ geographies. The study also highlights the need for OEMs to dynamically assess IP assertion risk, especially as new technologies attract non-traditional actors into the ecosystem.
This thesis contributes to the limited academic literature on patent country selection by offering a strategy oriented approach applicable to automotive OEMs. It emphasizes that country selection decisions should be aligned with the strategic role of each patent portfolio, rather than treated as an isolated administrative task
Biometric Authentication for the Web: A Face Recognition System
No full textThis bachelor thesis presents the design and implementation of a proof of concept web application for biometric authentication using face recognition. The goal was to investigate if face recognition could function as an accurate, user-friendly and secure alternative to password based login systems on web platforms. The project included developing a face recognition pipeline using existing open-source models, with added features such as anti-spoofing and encryption for data protection. The system was implemented as a web application and was evaluated through a set of user tests and performance tests on datasets. The results show that the system achieves a high accuracy and usability, even though spoofing remains as an issue. Future work includes improving the spoofing detection, fine-tuning the models for better generalization and developing the system into a scalable authentication API
Mapping invisible narratives for future imaginaries; in Nandipara, Dhaka and beyond
No full textUrban planning has long relied on maps to understand and manage spatial relations, yet mapping is
never neutral. It is a political act that shapes how cities are imagined, governed, and transformed, often
reinforcing dominant power structures while rendering informal realities invisible. This research investigates
the potential of mapping as a critical and participatory tool to challenge such narratives and support more
ecological, just, and inclusive urban transitions.
Focusing on the Jirani Canal area in Nandipara, a rapidly urbanizing peripheral neighborhood in East
Dhaka, the study examines how formal planning discourses overlook the everyday practices, spatial
adaptations, and ecological logics of informal urban development. Nandi Para embodies many of Dhaka’s
contemporary challenges, climate vulnerability, infrastructural deficits, and migratory pressures, yet also
offers grounded examples of community-managed public spaces, informal circulation networks, and
smallscale urban agriculture.
Using a critical mapping framework, the research adopts a mixed-methods approach combining GISbased
spatial analysis, historical cartographic review, participatory mapping workshops, observational
fieldwork, and speculative design. It engages with multiple actors, local residents, schoolchildren, farmers,
and planning professionals, to map and interpret the lived realities of the area. By analyzing contested
mappings, the study surfaces “invisible narratives”: ecological corridors, informal economies, and adaptive
urban strategies often omitted from official maps.
Rather than producing a fixed masterplan, the thesis embraces mapping as an open-ended, reflective,
and collective process that foregrounds contested space, plural knowledges, and future possibilities. The
research advocates for a more participatory and speculative modes of map-making that not only document
but also reimagine the city. In doing so, it contributes to a broader discourse on the role of mapping
in shaping urban futures, particularly in contexts marked by informality, rapid change, and institutional
fragmentatio