Beata Jungselius, Maja Fröjelin, Sebastian Johansson
This study offers a novel contribution by examining the micro-level dynamics of sharenting through a Nordic lens, focusing on how Swedish parents perceive, experience, and manage this increasingly common social media practice. Through thematic analysis of semi-structured interviews with twelve Swedish parents, we conceptualize the practice of sharing visual representations of children and family life on social media. The findings reveal complex and often ambiguous reasonings where parents balance a desire to document, share and interact with concerns about social expectations, integrity and future implications. While they articulate clear boundaries around what not to share (e.g., nudity, sensitive situations), their reasoning around what is acceptable to post tends to be more implicit and negotiated, pointing to unspoken norms and evolving community standards. Findings illustrate how parents navigate these dilemmas through a set of media ideologies, explicit and implicit idioms of practice developed within their communities, and influences from Nordic digital culture. By unpacking details of activities and nuanced strategies that parents employ, this study advances understanding of sharenting as a socially negotiated, culturally situated practice, contributing new insights into how digital parenting unfolds in the Nordic context. The findings highlight the need for digital platforms to support and guide responsible sharing practices that respect children’s integrity while also allowing for parental expression.
History of scholarship and learning. The humanities, Social Sciences
The universal learning framework has been developed to obtain guarantees on the learning rates that hold for any fixed distribution, which can be much faster than the ones uniformly hold over all the distributions. Given that the Empirical Risk Minimization (ERM) principle being fundamental in the PAC theory and ubiquitous in practical machine learning, the recent work of arXiv:2412.02810 studied the universal rates of ERM for binary classification under the realizable setting. However, the assumption of realizability is too restrictive to hold in practice. Indeed, the majority of the literature on universal learning has focused on the realizable case, leaving the non-realizable case barely explored. In this paper, we consider the problem of universal learning by ERM for binary classification under the agnostic setting, where the ''learning curve" reflects the decay of the excess risk as the sample size increases. We explore the possibilities of agnostic universal rates and reveal a compact trichotomy: there are three possible agnostic universal rates of ERM, being either $e^{-n}$, $o(n^{-1/2})$, or arbitrarily slow. We provide a complete characterization of which concept classes fall into each of these categories. Moreover, we also establish complete characterizations for the target-dependent universal rates as well as the Bayes-dependent universal rates.
We study contrastive learning under the PAC learning framework. While a series of recent works have shown statistical results for learning under contrastive loss, based either on the VC-dimension or Rademacher complexity, their algorithms are inherently inefficient or not implying PAC guarantees. In this paper, we consider contrastive learning of the fundamental concept of linear representations. Surprisingly, even under such basic setting, the existence of efficient PAC learners is largely open. We first show that the problem of contrastive PAC learning of linear representations is intractable to solve in general. We then show that it can be relaxed to a semi-definite program when the distance between contrastive samples is measured by the $\ell_2$-norm. We then establish generalization guarantees based on Rademacher complexity, and connect it to PAC guarantees under certain contrastive large-margin conditions. To the best of our knowledge, this is the first efficient PAC learning algorithm for contrastive learning.
Abstract A recent study, following a subtle but unusual for history of science argumentation method starting from the premises established by stylometry, discovered a drastic stylistic contrast between Copernicus’s early opus Commentariolus and his mature writings. The finding challenged the long-established view that Copernicus became a humanistically minded scholar early in his life and composed Commentariolus between 1509 and 1514. The present study verifies and extends this investigation by analyzing the whole extant literary oeuvre of Copernicus. The stylometric investigation for the most critical stylistic markers is followed by a detailed linguistic analysis. The results allow for convincing guesses on what motivated Copernicus to abruptly change his Latin style and establish the circumstances under which Copernicus most likely fully embraced Renaissance humanism. In addition, the subsequent historiographical analysis presents a plausible hypothesis on the precise dating of Commentariolus, the “birth certificate” of Copernicus’s geokinetic cosmology. Thus, the utility of stylometry for the history of science is demonstrated.
As the backbone of Malaysia’s economy, small and medium-sized enterprises (SMEs), particularly those in the manufacturing industry, are spotlighted in times of crisis, considering the substantial risks that small businesses must overcome. SME manufacturers strive to drive productivity and enhance competitiveness amid ongoing global challenges. Sustainable manufacturing performance is a key indicator in the manufacturing industry for reviving SMEs in the recovery stage of global crises. With continuous technological advancement, industry 4.0 competencies can have significant influence on expanding the evolution of contemporary manufacturing from theoretical industry 4.0 concepts to real-world applications. There is a gap in the literature regarding the comprehensive investigation of industry 4.0 competencies collectively and their direct impact on sustainable manufacturing performance, as well as the potential moderating influence of rational culture. The research question addressed in this paper aims to provide a comprehensive overview of prevailing studies on the relationship between industry 4.0 competencies and sustainable manufacturing performance. This study references the Global Reporting Initiative and identifies relevant industry 4.0 competencies, conducting a systematic literature review to investigate the relationship between industry 4.0 competencies and sustainable manufacturing performance and the role of rational culture as a moderating variable. Examining research published between 2013 and 2023, only 33 documents satisfy the eligibility criteria. This paper carries implications for SME manufacturers and government in Malaysia, highlighting the importance of recognizing and implementing appropriate industry 4.0 competencies to improve sustainable manufacturing performance and reap associated benefits.
History of scholarship and learning. The humanities, Social Sciences
Maria Rannielly de Araújo Lima Magalhães, Gabriel Huet Borges de Arruda , Cynthia Melo
A comunicação de más notícias (CMN) é um grande desafio para profissionais atuantes em unidades neonatais, cujo entraves precisam ser mais bem compreendidos. Objetivou-se descrever a experiência do processo de CMN e a capacitação para tal a partir da perspectiva dos profissionais de saúde em contexto neonatal. Realizou-se uma pesquisa qualitativa, com participação de 23 profissionais, que responderam a um roteiro de entrevista semiestruturado. Os dados foram analisados a partir do software IRaMuTeQ e de análise de conteúdo. Os resultados apresentaram que a maioria dos profissionais de saúde considera a CMN uma tarefa desafiadora no contexto neonatal; que envolve diferentes adversidades e não apenas a notícia de morte; e que eles não foram capacitados para esse processo, embora reconheçam a importância da preparação. Conclui-se a importância de formações continuadas para capacitar profissionais com habilidades técnicas e socioemocionais para a CMN, melhorando a qualidade da comunicação e relação com o paciente e família.
Education (General), History of scholarship and learning. The humanities
In recent years, the tourism industry has expanded rapidly and China will enter the mass tourism era comprehensively, with tourism becoming an important leisure lifestyle for people. However, natural disasters have the characteristics of high frequency, multiple types, wide distribution and large post-disaster losses, which often cause huge impact on tourism industry development. It is of great practical significance to explore the key factors for successful rescue of tourists after disasters. Taking the Wenchuan earthquake as an example, the article selects the Fuzzy Delphi Method and the Analytical Hierarchy Process as the data analysis method, and also randomly selects the local residents of 20 affected scenic spots as the research subjects. The results show that among the assessment dimensions, the most important dimension is prevention, followed by preparedness, response and recovery respectively. Meanwhile, among the 16 assessment indicators, the top five most valued indicators were regional emergency planning, management systems, prompt notification, publicity and education, training and emergency drill in order.
History of scholarship and learning. The humanities, Social Sciences
En los últimos años se han planteado varias críticas al modelo de orientalismo propuesto por Edward Said. Entre ellas destacan, por ejemplo, las advertencias acerca de la existencia de discursos orientalistas en fecha muy anterior a la señalada por Said o de corrientes orientalistas, como las ibéricas, que no encajan en el modelo antagonista propuesto por aquél. Este trabajo partirá de los textos sobre el Sáhara argelino elaborados por Yves Alliaume en las décadas centrales del siglo XX para ofrecer una reinterpretación de la relación entre los discursos orientalistas y la práctica de la colonialidad en Argelia. Para ello, se centrará en las nociones de lo “bereber” y del estatus del islam entre las poblaciones locales para mostrar como la caracterización de las sociedades en el periodo colonial no partía de unas distinciones culturales netamente definidas, sino que se articulaba a través de valoraciones sociales que conectan el ámbito colonial con el metropolitano. De todo ello surge una dimensión de la antropología del norte de África raramente advertida hoy día, pero que ya fue señalada en su momento por Franz Fanon y que aparece recurrentemente en los textos del Yves Alliaume. Esta relectura social puede servir para recomponer la noción del orientalismo dentro de la noción foucaultiana de discurso que la inspiró, así como para explorar su compleja relación con la práctica de la colonialidad desde una perspectiva glocal.
History of scholarship and learning. The humanities, Language and Literature
Abstract The research aims to assess the water quality of the Tharthar Canal and its suitability for human activities in the study area by conducting physical and chemical analyzes of that water and showing its suitability for human uses. The results of the study show that most of the samples in the study area are not suitable for human use, especially drinking. The total hardness, sulfate and calcium values were determined only after filtering and processing to become drinkable, while laboratory analyzes showed that they are suitable for agricultural use in all samples of the study area. Because of the importance of the research, the physical and chemical properties were addressed to find out if they were affected by the natural or human geographical characteristics and their impact on the residents of the nearby areas and their various activities to find out the problems and develop appropriate solutions that would address the problems and adopt it as a fixed water source. It became clear through the field study that the Al-Tharthar-Euphrates Canal is not approved by the population as a main water resource, but most of the residents of the areas relied on wells for their various activities, foremost of which is agriculture and drinking.
History of scholarship and learning. The humanities
Deep neural networks can yield good performance on various tasks but often require large amounts of data to train them. Meta-learning received considerable attention as one approach to improve the generalization of these networks from a limited amount of data. Whilst meta-learning techniques have been observed to be successful at this in various scenarios, recent results suggest that when evaluated on tasks from a different data distribution than the one used for training, a baseline that simply finetunes a pre-trained network may be more effective than more complicated meta-learning techniques such as MAML, which is one of the most popular meta-learning techniques. This is surprising as the learning behaviour of MAML mimics that of finetuning: both rely on re-using learned features. We investigate the observed performance differences between finetuning, MAML, and another meta-learning technique called Reptile, and show that MAML and Reptile specialize for fast adaptation in low-data regimes of similar data distribution as the one used for training. Our findings show that both the output layer and the noisy training conditions induced by data scarcity play important roles in facilitating this specialization for MAML. Lastly, we show that the pre-trained features as obtained by the finetuning baseline are more diverse and discriminative than those learned by MAML and Reptile. Due to this lack of diversity and distribution specialization, MAML and Reptile may fail to generalize to out-of-distribution tasks whereas finetuning can fall back on the diversity of the learned features.
Martin Gauch, Maximilian Beck, Thomas Adler
et al.
We introduce SubGD, a novel few-shot learning method which is based on the recent finding that stochastic gradient descent updates tend to live in a low-dimensional parameter subspace. In experimental and theoretical analyses, we show that models confined to a suitable predefined subspace generalize well for few-shot learning. A suitable subspace fulfills three criteria across the given tasks: it (a) allows to reduce the training error by gradient flow, (b) leads to models that generalize well, and (c) can be identified by stochastic gradient descent. SubGD identifies these subspaces from an eigendecomposition of the auto-correlation matrix of update directions across different tasks. Demonstrably, we can identify low-dimensional suitable subspaces for few-shot learning of dynamical systems, which have varying properties described by one or few parameters of the analytical system description. Such systems are ubiquitous among real-world applications in science and engineering. We experimentally corroborate the advantages of SubGD on three distinct dynamical systems problem settings, significantly outperforming popular few-shot learning methods both in terms of sample efficiency and performance.
Krunoslav Lehman Pavasovic, Jonas Rothfuss, Andreas Krause
Meta-learning aims to extract useful inductive biases from a set of related datasets. In Bayesian meta-learning, this is typically achieved by constructing a prior distribution over neural network parameters. However, specifying families of computationally viable prior distributions over the high-dimensional neural network parameters is difficult. As a result, existing approaches resort to meta-learning restrictive diagonal Gaussian priors, severely limiting their expressiveness and performance. To circumvent these issues, we approach meta-learning through the lens of functional Bayesian neural network inference, which views the prior as a stochastic process and performs inference in the function space. Specifically, we view the meta-training tasks as samples from the data-generating process and formalize meta-learning as empirically estimating the law of this stochastic process. Our approach can seamlessly acquire and represent complex prior knowledge by meta-learning the score function of the data-generating process marginals instead of parameter space priors. In a comprehensive benchmark, we demonstrate that our method achieves state-of-the-art performance in terms of predictive accuracy and substantial improvements in the quality of uncertainty estimates.
Louis Kirsch, James Harrison, Jascha Sohl-Dickstein
et al.
Modern machine learning requires system designers to specify aspects of the learning pipeline, such as losses, architectures, and optimizers. Meta-learning, or learning-to-learn, instead aims to learn those aspects, and promises to unlock greater capabilities with less manual effort. One particularly ambitious goal of meta-learning is to train general-purpose in-context learning algorithms from scratch, using only black-box models with minimal inductive bias. Such a model takes in training data, and produces test-set predictions across a wide range of problems, without any explicit definition of an inference model, training loss, or optimization algorithm. In this paper we show that Transformers and other black-box models can be meta-trained to act as general-purpose in-context learners. We characterize transitions between algorithms that generalize, algorithms that memorize, and algorithms that fail to meta-train at all, induced by changes in model size, number of tasks, and meta-optimization. We further show that the capabilities of meta-trained algorithms are bottlenecked by the accessible state size (memory) determining the next prediction, unlike standard models which are thought to be bottlenecked by parameter count. Finally, we propose practical interventions such as biasing the training distribution that improve the meta-training and meta-generalization of general-purpose in-context learning algorithms.
El artículo expone los resultados de una investigación centrada en el estudio de las trayectorias de los guatemaltecos que, apoyados por el Programa de Becas de la Fundación Rockefeller, cursaron capacitaciones avanzadas o estudios de posgrado en ciencias agrícolas en México, entre 1949 y 1976. Desde el mirador de la historia del intercambio académico, en clave transnacional, se propone que la agricultura fungió como un puente que, a través de los itinerarios de los becarios, unió dos procesos: la profesionalización de la enseñanza agrícola en México y la profesionalización de la investigación agrícola en Guatemala. Se argumenta, además, que ambos procesos formaron parte del despliegue de la revolución verde en América Latina. Esta indagación se fundamentó en la consulta de fuentes primarias, hemerografía y entrevistas.
History of scholarship and learning. The humanities, History (General) and history of Europe
صحيحٌ أنّ الحياة تجدد، والتّجدد سرّ الوجود، وصحيحٌ أنّ الوجود لو غاسسحب عنه التّجدد فارق الحياة، لكنّ الماضي أيضا له عبقه وحيويته ووجوده الكائن فيه، وما يحتاجه تجدد فقط ليحيا حياة ثانية، ولكيلا يَهمد ويندثر، تهدف هذه الدّراسة إلى تجدد التّراث، والتّعرف على علم الاكتناه وببلوغرافية المخطوط وأثرهما في إثراء السّاحة العلميّة والثّقافيّة، يتيح هذا المفهوم بما هو علم وفنّ فرصة الكشف عن تماثل إبداعيّ في شتّى العلوم والفنونّ، بين سالف العصور وبين الواقع الحاليّ، وأثبتت الدّراسة أنّ التّراث العربيّ الإسلاميّ عبارة عن بِساطٍ زاهٍ ثريّ، نسجته أيد وأفكار مبدعة معطاء، أفرغت طاقتها وثقافتها وأدبها فيه، وكلّ تلك القرائح المُبدعة تستحق منّا الإعزاز والإكرام، فكأين من تجربة وعلم وإبداع هما طيّ للنّسيان لولا المعول على تراث الأجداد، إذ من خلاله يمكن فهم العلائق بين ما كان وما يكون، وللعرب المشرقيين يُعزى فضل إنجاب المبدعين في شتّى المجالات، ولهم قصب السّبق للإنتاج الإبداعيّ، فهمّ أمّة ولود متنوّرة خلّفت أعظم الهامات، وأسست أهمّ العلوم كذلك الفنون، وهذا ما يُثير فينا الإعجاب والتّقصي لدراسته، والنّاظر لمسار اللّغة العربيّة وأثَر مواقع وقنوات التّواصل الثّقافيّ و(التكنلوجيا) الحديثة يرى بجلاء الخطر الماحق، والمصير المُرعب الّذي قد تنحدر إليه ثقافتنا، الّذي يجعل مصيبتنا في تراثنا ولغتنا تتصاغر دونها كلّ مصيبة، ولابدّ إذا من تضافر جهود الغيارى على التّراث الثّقافيّ والدّينيّ واللّغويّ، فيعمدون لإحياء الكتب الصّفراء الغنيّة المغمورة، والاهتمام بها، ففيها علم وفنّ زاخران، هذا وقد اعتمدنا المنهج الوصفيّ وهو ما يناسبنا في بيان أهميّة التّحقيق في رفد الثّقافة، وأبنّا مدى تأثير عمل المحقِّق سِلبا وإيجابا في إثراء التّراث.
History of scholarship and learning. The humanities, Arts in general
The COVID-19 pandemic has impacted almost every aspect of life. According to data from several countries, book reading has surged since the beginning of the imposed lockdowns. This situation has presented cultural mediators with an unprecedented opportunity to influence their audiences’ reading habits. The current study reports the results of a qualitative analysis of Iranian news websites’ book suggestions during March 2020, the peak of the first wave of COVID-19 in Iran. Through analysing the suggested books’ original language, publication date, genre, and publisher, the study uncovers some of the politics of book promotion in Iran. The findings revealed a high level of homogeneity among the reading suggestions in terms of original language, genre, and publishers, raising the possibility that wider government policies encourage the promotion or marginalisation of certain types of books. Fiction and nonfiction books about the Iran-Iraq War (1980–1988) dominated the suggestions, while comedy and nonpolitical mystery/thrillers were almost nonexistent. These findings and more are discussed in light of the sociopolitical context in Iran.
History of scholarship and learning. The humanities, Social sciences (General)
Following the success of deep learning in a wide range of applications, neural network-based machine-learning techniques have received significant interest for accelerating magnetic resonance imaging (MRI) acquisition and reconstruction strategies. A number of ideas inspired by deep learning techniques for computer vision and image processing have been successfully applied to nonlinear image reconstruction in the spirit of compressed sensing for accelerated MRI. Given the rapidly growing nature of the field, it is imperative to consolidate and summarize the large number of deep learning methods that have been reported in the literature, to obtain a better understanding of the field in general. This article provides an overview of the recent developments in neural-network based approaches that have been proposed specifically for improving parallel imaging. A general background and introduction to parallel MRI is also given from a classical view of k-space based reconstruction methods. Image domain based techniques that introduce improved regularizers are covered along with k-space based methods which focus on better interpolation strategies using neural networks. While the field is rapidly evolving with plenty of papers published each year, in this review, we attempt to cover broad categories of methods that have shown good performance on publicly available data sets. Limitations and open problems are also discussed and recent efforts for producing open data sets and benchmarks for the community are examined.