Publikationer från Linköpings universitet
Not a member yet
58266 research outputs found
Sort by
Finding Minimum-Cost Explanations for Predictions made by Tree Ensembles
No full textThe ability to reliably explain why a machine learning model arrives at a particular prediction is crucial when used as decision support by human operators of critical systems. The provided explanations must be provably correct, and preferably without redundant information, called minimal explanations. In this paper, we aim at finding explanations for predictions made by tree ensembles that are not only minimal, but also minimum with respect to a cost function. To this end, we first present a highly efficient oracle that can determine the correctness of explanations, surpassing the runtime performance of current state-of-the-art alternatives by several orders of magnitude. Secondly, we adapt an algorithm called MARCO from the optimization field (calling the adaptation m-MARCO) to compute a single minimum explanation per prediction, and demonstrate an overall speedup factor of 2.7 compared to a state-of-the-art algorithm based on minimum hitting sets, and a speedup factor of 27 compared to the standard MARCO algorithm which enumerates all minimal explanations.This is a pre-print publication also published at ArXiv, which has not been under subject to peer-review.</p
Dirac Nodal Arc Semimetal PtSn4: An Ideal Platform for Understanding Surface Properties and Catalysis for Hydrogen Evolution
No full textConductivity, carrier mobility, and a suitable Gibbs free energy are important criteria that determine the performance of catalysts for a hydrogen evolution reaction (HER). However, it is a challenge to combine these factors into a single compound. Herein, we discover a superior electrocatalyst for a HER in the recently identified Dirac nodal arc semimetal PtSn4. The determined turnover frequency (TOF) for each active site of PtSn4 is 1.54 H-2 s(-1) at 100 mV. This sets a benchmark for HER catalysis on Pt-based noble metals and earth-abundant metal catalysts. We make use of the robust surface states of PtSn4 as their electrons can be transferred to the adsorbed hydrogen atoms in the catalytic process more efficiently. In addition, PtSn4 displays excellent chemical and electrochemical stabilities after long-term exposure in air and long-time HER stability tests.Funding Agencies|European Research Council (ERC Advanced Grant) [291472]; ERC Advanced Grant [742068]</p
Older couples with long-term disabilities : Multiple jeopardy or successful ageing?
No full textIn a study where older couples who had lived long lives with physical disabilities were interviewed about mutual care, an unexpected finding that while the couples reported worsening disabilities, they also described themselves as advantaged, and as maintaining a high level of activity in some areas of life. In this article, we investigate how the couples achieve this sense of ageing successfully when the participants’ stories of physical decline, from an outsider’s perspective, might be thought to produce double-, triple- or multiple jeopardy. The results show how the couples use different types of reference groups and refer to various resources to frame themselves as privileged. Furthermore, through environmental adaptations, technical aids and supportive services, they managed to maintain a high level of activity, despite living with disabilities. Notably, the couples described themselves as advantaged by referring to being older, to the long life with disabilities and to ageing with disabilities together. The overarching ambition of the article is to problematize established notions of successful ageing. Considering that the couples referred to the very conditions that might be seen as disadvantageous, the problem of pre-defined criteria for successful ageing becomes particularly apparent.This manuscript will not be published.</p
Measuring physical activity with activity monitors in patients with heart failure: from literature to practice. A position paper from the Committee on Exercise Physiology and Training of the Heart Failure Association of the European Society of Cardiology
No full textThe aims of this paper were to provide an overview of available activity monitors used in research in patients with heart failure and to identify the key criteria in the selection of the most appropriate activity monitor for collecting, reporting, and analysing physical activity in heart failure research. This study was conducted in three parts. First, the literature was systematically reviewed to identify physical activity concepts and activity monitors used in heart failure research. Second, an additional scoping literature search for validation of these activity monitors was conducted. Third, the most appropriate criteria in the selection of activity monitors were identified. Nine activity monitors were evaluated in terms of size, weight, placement, costs, data storage, water resistance, outcomes and validation, and cut-off points for physical activity intensity levels were discussed. The choice of a monitor should depend on the research aims, study population and design regarding physical activity. If the aim is to motivate patients to be active or set goals, a less rigorously tested tool can be considered. On the other hand, if the aim is to measure physical activity and its changes over time or following treatment adjustment, it is important to choose a valid activity monitor with a storage and battery longevity of at least one week. The device should provide raw data and valid cut-off points should be chosen for analysing physical activity intensity levels. Other considerations in choosing an activity monitor should include data storage location and ownership and the upfront costs of the device.Funding Agencies|European Society of Cardiology Funding Source: Medline</p
Towards an equitable future of global photovoltaic waste recycling
No full textThe world is facing a mounting challenge from the rapid increase in global photovoltaic (PV) waste, as the first generation of PV systems has reached their service life1,2. The climate and economic performance of PV waste management depends heavily on the selection of appropriate recycling strategies3,4. Here, we develop a comprehensive framework to evaluate recycling strategies across 30 regions from 2020 to 2060, assessing the economic and climate benefits of local versus outsourced recycling, covering major recycling technologies. Global PV waste is expected to reach 175–305 million tons (Mt) by 2060, with middle-income countries like China becoming major contributors post-2040. Recycling can reduce greenhouse gas emissions by up to 117 Mt CO2 equivalent and yield US$3.3–18.9 billion in annual net benefits by 2060. Outsourced recycling lowers costs, however, raises inequity concerns for low-income regions. Our results highlight that subsidies temporarily mitigate disparities however wane in long-term efficacy. We suggest that phased subsidy frameworks and international cooperation, with a focus on technology transfer and funding for recycling capacity in low-income countries, provide effective ways to achieve equitable and scalable PV waste circularity.This is a preprint in Research Square; it has not been peer reviewed by a journal.Version 1 posted 27 May, 2025</p
Time Series Fault Classification for Wave Propagation Systems with Sparse Fault Data
No full textIn this work Time Series Classi cation techniques are investigated, and especially their applicability in applications where there are signi cant di erences between the individuals where data is collected, and the individuals where the classi cation is evaluated. Classi cation methods are applied to a fault classi cation case, where a key assumption is that data from a fault free reference case for each speci c individual is available. For the investigated application, wave propagation cause almost chaotic changes of a measured pressure signal, and physical modeling is di cult. Direct application of One-Nearest-Neighbor Dynamic Time Warping, a common technique for this kind of problem, and other machine learning techniques are shown to fail for this case and new methods to improve the situation are presented. By using relative features describing the di erence from the reference case rather than the absolute time series, improvements are made compared to state-of-the-art time series classi cation algorithms.This is a preprint, arXiv:2203.16121, posted on ArXiv. The fulltext was made available on ArXiv on Wed, 30 Mar 2022 08:02:26 UTC and with licence CC BY-NC-ND 4.0. The preprint has not been formally peer-reviewed by ArXiv.</p