Repository for Publications and Research Data
News from the ETH Library
Recently Added
Balanced Sharpness-Aware Minimization for Imbalanced Regression
Item type: Conference Paper
Liu, Yahao; Wang, Qin; Duan, Lixin; et al. (2025)
Regression is fundamental in computer vision and is widely used in various tasks including age estimation, depth estimation, target localization, etc. However, real-world data often exhibits imbalanced distribution, making regression models perform poorly especially for target values with rare observations (known as the imbalanced regression problem). In this paper, we reframe imbalanced regression as an imbalanced generalization problem. To tackle that, we look into the loss sharpness property for measuring the generalization ability of regression models in the observation space. Namely, given a certain perturbation on the model parameters, we check how model performance changes according to the loss values of different target observations. We propose a simple yet effective approach called Balanced Sharpness-Aware Minimization (BSAM) to enforce the uniform generalization ability of regression models for the entire observation space. In particular, we start from the traditional sharpness-aware minimization and then introduce a novel targeted reweighting strategy to homogenize the generalization ability across the observation space, which guarantees a theoretical generalization bound. Extensive experiments on multiple vision regression tasks, including age and depth estimation, demonstrate that our BSAM method consistently outperforms existing approaches. The code is available here.
Supply chain antifragility capabilities: a systematic literature network analysis and conceptual framework
Item type: Review Article
Inyo, Doreen N.; Wagner, Stephan M.; Githii, Michael Wainaina; et al. (2026)
Purpose - Supply chains are essential to economic growth in many industries. However, their complexity and slow response to market changes leave them vulnerable to environmental disruptions. The concept of antifragility explains how organizations can withstand and even thrive during disruptions. The goal of this paper is to review and consolidate the literature on supply chain antifragility and derive a conceptual framework that illustrates how antifragility-oriented capabilities can transform disruptions into long-lasting performance improvements.
Design/methodology/approach - A systematic literature network analysis (SLNA) and a keyword analysis were conducted on articles related to antifragility and supply chain capabilities published in the Scopus database between 2012 and June 2025. Based on the resulting thematic clusters identified through SLNA and keyword analysis, additional literature, and dynamic capability theory, a novel conceptual framework for supply chain antifragility was developed.
Findings - The SLNA reveals three principal thematic clusters: supply chain vulnerabilities, supply chain enablers, and supply chain capabilities for antifragility. These thematic clusters converge in an integrative conceptual framework showing how antifragility emerges from interactions among disruption drivers and triggers, technological and informational enablers, and antifragility capabilities. These factors collectively facilitate enhanced supply chain antifragility outcomes.
Originality/value - Introduced in 2012, antifragility offers an alternative perspective to the traditional concepts of robustness and resilience. It has only recently been applied to supply chains, which are subject to common disorder and disruption. By examining the concept in this context and reviewing relevant work, the analyses and conceptual framework developed in this study provide a comprehensive overview, establish a critical foundation for future research, and inform how to take advantage of supply chain antifragility.
A neuromorphic continuous soil monitoring system for precision irrigation
Item type: Conference Paper
Tincani, Mirco; Kerouch, Khaled; Garlando, Umberto; et al. (2025)
Sensory processing at the edge requires ultra-low power stand-alone computing technologies. This is particularly true for modern agriculture and precision irrigation systems which aim to optimize water usage by monitoring key environmental observables continuously using distributed efficient embedded processing elements. Neuromorphic processing systems are emerging as a promising technology for extreme edgecomputing applications that need to run on resource-constrained hardware. As such, they are a very good candidate for implementing efficient water management systems based on data measured from soil and plants, across large fields. In this work, we present a fully energy-efficient neuromorphic irrigation control system that operates autonomously without any need for data transmission or remote processing. Leveraging the properties of a biologically realistic spiking neural network, our system performs computation, and decision-making locally. We validate this approach using real-world soil moisture data from apple and kiwi orchards applied to a mixed-signal neuromorphic processor, and show that the generated irrigation commands closely match those derived from conventional methods across different soil depths. Our results show that local neuromorphic inference can maintain decision accuracy, paving the way for autonomous, sustainable irrigation solutions at scale.
The health production function in South Asia: panel evidence on education, health expenditure, governance, and longevity
Item type: Journal Article
Ridwan, Mohammad; Antor, Zulfiquar Ali; Ko, Jeremy; et al. (2026)
This study examines the association of education, government effectiveness, health expenditure, economic growth, and urbanization with life expectancy in South Asia within Grossman’s health capital framework. Using annual panel data for seven South Asian countries—Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan, and Sri Lanka—over 2000–2025, the analysis investigates both long-run and short-run relationships. The empirical strategy combines panel unit root tests, panel ARDL estimation, Driscoll–Kraay standard errors, PCSE, FGLS, System GMM, and bootstrap panel causality tests. The results indicate a stable long-run relationship among the variables. Education, measured by school enrollment and adult literacy, is positively and significantly associated with life expectancy. GDP per capita also shows a positive and significant association. Health expenditure has a positive but model-sensitive association, suggesting that the effectiveness of spending may matter more than its level alone. By contrast, government effectiveness and urbanization are significantly associated with life expectancy but with negative coefficients in the estimated models. Bootstrap causality results indicate dynamic links between life expectancy and several explanatory variables. Overall, the findings highlight the importance of education, inclusive growth, efficient health spending, and better-managed urban development in shaping longevity across South Asia.
Ketene, Diketene, and 1,3-Cyclobutanedione Chemistry-Industrial Relevance, Reactivity, and Emerging Applications of Cyclobutane Motifs
Item type: Journal Article
Kohler, Philipp; Taeschler, Christoph; Stepen, Arne Jan; et al. (2026)
Ketene and diketene are cornerstone intermediates of industrial chemistry, enabling synthesis of a wide range of functional products. Beyond their established applications, side products of ketene dimerization provide access to 1,3-cyclobutanedione and triketene chemistry. This review highlights industrial relevance, reactivity, scalable access to 1,3-cyclobutanedione via the dione salt, and emerging applications of cyclobutane motifs in pharmaceuticals, emphasizing opportunities for further derivatization and process development.