Publikationer från Umeå universitet
Not a member yet
55455 research outputs found
Sort by
Liquid Neural Networks in Finance : An Application of Machine Learning for Anomaly detection in Trading
No full textAnomaly detection in financial transactions is essential for identifying irregular trading behaviors and mitigating financial risks. This is important because trades that deviate from expected patterns can expose financial institutions to financial, regulatory, or reputational consequences. As modern financial markets grow increasingly complex, traditional monitoring methods struggle to adapt to rapidly changing patterns in large-scale trading data. This thesis investigates and compares the performance of Liquid Neural Networks (LNN) and Long Short-Term Memory (LSTM) networks for predictive modeling and anomaly detection in financial trading data. LNNs, which are known for their ability to process fast-changing time-series inputs, are being trained to predict a trader’s next transaction based on their historical behavior. Anomalies are identified by deviations from these predictions. The performance of LNNs is benchmarked against LSTM models using transaction data provided by Handelsbanken and evaluated under two test conditions, one using the original dataset and the other being an altered dataset with synthetically inserted anomalies. Two modeling approaches are applied, a predictive model and an autoencoder-based model. Both approaches serve to assess the models’ ability to capture underlying behavioral patterns and detect anomalies. In predictive modeling, the LNN generally outperforms the LSTM across most numerical and categorical metrics when evaluated on the original dataset, indicating a stronger ability to model normal trading behavior. However, this comes at the cost of significantly longer training times. The LSTM, by contrast, is more computationally efficient and trains faster, making it a practical choice despite its larger architecture. When detecting synthetically inserted anomalies, both models perform closely across the predictive and autoencoder-based architectures. The LSTM exhibits a slight edge over the LNN in terms of anomaly detection performance. Overall, both models demonstrate strengths and limitations depending on the task. The LNN excels in predictive modeling of normal trading behavior, while the LSTM provides faster training and better overall performance when anomalies are present. If successful, this approach can enhance the effectiveness of financial surveillance systems and support anomaly detection strategies
“The system brought beauty to our community” : Evaluating the impact of a physical address system in Mathare informal settlement, Nairobi, through ripple effect mapping
No full textIntroduction: Rapid urbanisation in Nairobi has led to the growth of informal settlements, characterised by overcrowding, poor infrastructure, and limited access to basic services. Often invisible in formal systems, these areas face weak accountability and restricted service access, worsening health and wellbeing. This paper evaluates the impact of a community-led Physical Address System (PAS) in improving visibility, access, and support for residents in the settlement of Kiamutisya in Mathare, Nairobi. Methods: Drawing on community-led data collection, the PAS introduced a unique address for each physical structure in Kiamutisya to support service delivery and emergency response, addressing the invisibility that residents face. We held three Ripple Effect Mapping (REM) workshops with community leaders, community health promoters and Accountability and Responsiveness in Informal Settlements for Equity (ARISE) Hub co-researchers. REM is a participatory evaluation method that captures the impacts of an intervention, both intended and unintended, from the perspective of the community. Data was coded using NVivo 12 and analysed thematically. Results: The PAS evaluation found direct impacts like better service delivery by community health promoters, and indirect benefits including stronger community identity, improved safety, and government engagement. It also noted ethical risks like data misuse and resident stigmatization. Key recommendations for policymakers and planners include using physical addresses to enhance service delivery and emergency response, formally integrating addresses into planning systems, and fostering community participation to improve equity in challenging urban environments. Conclusions: This study underscores the potential of community health systems to address structural vulnerabilities and build resilience in informal settlements. The PAS demonstrates the value of participatory approaches in creating sustainable, scalable models for urban health, reinforcing the critical role of communities as active agents in advancing health and well-being
Geek nostalgia : the reflective and restorative defence of white male geek culture
No full textDuring recent decades, geek culture has become increasingly visible, and the geek has left the cultural margins, becoming more popular than ever. At the same time, nostalgia has emerged as a central component of geek culture. Framed by a post-structural understanding of gender and race and drawing on cultural theorist Svetlana Boym’s distinction between reflective and restorative nostalgia, this article explores how and why geeks nostalgically long for a time when they were largely marginalized. We combine readings of Swedish online geek podcasts and YouTube channels with ethnographic visits to geek conferences and pop-cultural “geek fairs,” such as Comic Con and SciFiWorld. We argue that geek nostalgia represents a clinging on to a “constitutive wound,” allowing the geek figure to mobilize masculine victimhood in ways that simultaneously underpin geek privilege and allow the geek to continue operating as a white male gatekeeper of geek culture
Urinary silicon excretion in relation to lactation and bone mineral density : a longitudinal study post-partum
No full textSilicon (Si) may be a mineral beneficial for bone health. Pregnancy and lactation have major impacts on maternal bone metabolism as bone minerals, including calcium (Ca), are required for growth of the foetus and for milk production. Like urinary Ca excretion, Si excretion has been reported to be high in pregnant women, but there are no data post-partum and during lactation. The aim of the present study was to investigate the urinary excretion of Si (U-Si), from the third trimester of pregnancy until 18 months post-partum, and in relation to the length of lactation, to determine if changes in U-Si are associated with changes in areal bone mineral density (aBMD). This longitudinal study included 81 pregnant women, of whom 56 completed the study. Spot urine samples were collected at the third trimester and at 0.5, 4, 12, and 18 months post-partum and were analysed for Si and Ca by ICP-OES. The aBMD was measured post-partum at lumbar spine and femoral neck by dual-energy x-ray absorptiometry. Women lactating for 4–8.9 and ≥ 9 months had significantly higher U-Si at 4 months post-partum, compared with the third trimester. No significant longitudinal differences in U-Si were found after correcting for creatinine. Changes in U-Si and in aBMD were not correlated, except at the lumbar spine from 0.5 to 12 months post-partum in the women lactating for 4–8.9 months. Taken together, our results suggest that there is a possibility that U-Si increases post-partum in women lactating for 4 months or longer, although it is not related to changes in aBMD
"Being a girl has become a public health problem!" : Feminist organizations both resisting and reproducing postfeminist mental health discourses
No full textAn apparent decline in youth mental health in the Nordic countries has received a lot of attention over the last decade, and young women have been described as a particularly vulnerable group. To combat the problem, political debate has centred on solutions at the individual level, despite significant evidence of the associations between structural factors and adverse mental health. In this article, I explore how two feminist support organizations working for young women in Sweden (tjejjourer) frame and address girls' mental health as a problem of feminist concern. Interviews with staff, organizational documents, websites, social media, and other communications were analysed following Snow and Benford's theory on collective action frames, to identify the organizations' problem diagnoses, solutions, and motivational frames for action. The results show that the organizations both reiterate and resist a postfeminist discourse on mental health. There is notable friction between structural problem diagnoses, based on intersectional analyses of power, and political demands and support oriented towards the individual. In conclusion, I argue that this friction may be a result of the organizations' adoption of discourses on risk, vulnerability, and mental health itself, but also that the complex relationship between structural critique and individual care is indispensable for feminist organizing
Provoked vulvodynia from a patient perspective : physiotherapy made a difference
No full textPurpose: Vulvodynia is considered to be a common cause of sexual pain in women of reproductive age and has a significant negative impact on their psycho-sexual health and quality of life. This study aimed to investigate the felt and known experience of living with provoked vulvodynia (PVD) in a group of women in Sweden and to explore the support, information, and treatment perceived to be important based on experienced symptoms. Methods: Ten women recruited by staff, from the vulva clinic at two hospitals in Sweden, participated in individual interviews. The results were analysed using qualitative content analysis. Results: The analysis resulted in the overarching theme ‘The women’s dilemmas regarding a sustainable daily life’. This theme is based on the difficulties the women experienced in being listened to and getting information and treatment to have the quality of life they want. Most important was understanding their own body, understanding the purpose of the treatment, and getting manual guidance to break the fear of pain. Conclusion: The results give a detailed picture of women’s experiences of PVD and delineate components of treatment perceived as important. This interview study is significant for healthcare professionals involved as the knowledge can contribute to faster diagnosis and better patient-specific treatment. The study may also guide future healthcare-related political decisions and the patient flow for these patients
The impact of peacekeeping on post-deployment earnings for Swedish veterans
No full textWe study the effect of peacekeeping on post-deployment earnings for military veterans. Using Swedish administrative data, we follow a sample of more than 11,000 veterans who were deployed for the first time during the period 1993-2010 for up to nine years after returning home. To deal with selection bias, we use difference-in-differences propensity score matching based on a rich set of covariates, including measures of individual ability, health and pre-deployment labour market attachment. We find that, overall, veterans’ long run post-deployment earnings are largely unaffected by their service. As such, our findings challenge the notion that Swedish veterans struggle on the labour market. At the same time, we find indications that the zero effect for the full sample hides interesting patterns in the effects from deployment, across sub-groups of veterans. Veterans deployed to Bosnia in the 1990s appear to have suffered a transitory earnings penalty, whereas veterans deployed to Afghanistan in the 2000s, as well as veterans with below average cognitive ability, and veterans with low-educated parents, appear to have earned a long run earnings premium
The impact of a soft robotic glove on activity performance and body function for brachial plexus birth injury patients : a longitudinal case series
No full textBackground: A brachial plexus birth injury (BPBI) can cause reduced ability to use the arm and hand in daily activities due to reduced grip strength and endurance. A soft robotic glove can increase the number of activities performed and improve activity performance for patients with neurological disease. The use of a soft robotic glove for patients with BPBI has not been studied. Purpose: To investigate if a soft robotic glove can improve activity performance and body function for patients with BPBI. Study design: Longitudinal Case Series. Methods: A convenience sample of patients with BPBI, treated by the Brachial plexus injury service in Umeå, Sweden were studied. Eight patients used a soft robotic glove, (Carbonhand®), at home for three months. Data on activity performance and satisfaction with activity performance, active range of motion and strength were collected at baseline, and at three and four months. A patient evaluation form was filled out at three months, all patients kept a diary for three out of 12 weeks. Results: Six out of eight patients wanted to continue using the device and improved their self-perception of activity performance and satisfaction with the performance due to a more secure grip, compared to when not using the device. All patients had improved maximum strength and endurance in elbow flexion at three months. The device was useful as an assisting device and as a training tool. Conclusion: A soft robotic glove (Carbonhand) may improve activity performance and perceived satisfaction and increase the number of activities that a person with BPBI can perform in everyday life. It is possible to increase strength in elbow flexion after using such a device. Due to this limited material, more research is needed