Hasil untuk "Art"

Jordan Hoffmann, Sebastian Borgeaud, Arthur Mensch et al.

We investigate the optimal model size and number of tokens for training a transformer language model under a given compute budget. We find that current large language models are significantly undertrained, a consequence of the recent focus on scaling language models whilst keeping the amount of training data constant. By training over 400 language models ranging from 70 million to over 16 billion parameters on 5 to 500 billion tokens, we find that for compute-optimal training, the model size and the number of training tokens should be scaled equally: for every doubling of model size the number of training tokens should also be doubled. We test this hypothesis by training a predicted compute-optimal model, Chinchilla, that uses the same compute budget as Gopher but with 70B parameters and 4$\times$ more more data. Chinchilla uniformly and significantly outperforms Gopher (280B), GPT-3 (175B), Jurassic-1 (178B), and Megatron-Turing NLG (530B) on a large range of downstream evaluation tasks. This also means that Chinchilla uses substantially less compute for fine-tuning and inference, greatly facilitating downstream usage. As a highlight, Chinchilla reaches a state-of-the-art average accuracy of 67.5% on the MMLU benchmark, greater than a 7% improvement over Gopher.

C. Danel, R. Moh, D. Gabillard et al.

M. Thompson, J. Aberg, H. Günthard

K. Paranjape

Paul Scowen, Manuel Quijada, Emily Kan et al.

In preparation for development of both key technologies and instrument concept studies to use those technologies, the Habitable Worlds Observatory Technology Maturation Project Office at the NASA Goddard Space Flight Center has compiled a series of statements of state of the art for those same key technologies. These statements are being provided to the public as exemplars and suggestions for possible future collaboration for those same instrument concept studies, but without mandate, to enable proposing teams to be able to find the technical solutions they need to field a compelling proposal. This information resides in the public domain and is presented without prejudice.

Eva Jarošová

The paper discusses the sepulchral art produced in 1500–1650 in Bohemia, especially in Prague, analysing the iconography of secular nobles and burghers in the context of early modern funerary sculpture. It explores the changes in funerary iconography, which depended on the deceased’s social status and profession. It also pays special attention to the depiction of women and children on tombstones, which reflected not only aesthetic norms but also religious and social conventions. It shows that tomb sculpture functioned not only as a memorial artefact, but also as a medium for self-presentation and visual communication that provided legitimacy to family claims and reflected contemporary conceptions of virtue, power, and eschatology.

Marina Beléndez, Lawrence Fisher

Background: Type 1 diabetes is a chronic condition that presents significant challenges not only for affected children and adolescents but also for their parents. The study examined the psychometric properties of the Spanish version of the Parent Diabetes Distress Scale (PDDS-SP) for parents of children and adolescents with type 1 diabetes. Method: Data were collected on 314 parents recruited through diabetes associations and Facebook groups. Participants completed the PDDS-SP, a well-being measure, and provided information about their child’s diabetes and treatment. Results: Exploratory factor analyses identified a 16-item, four-factor structure: parent-child relationship concerns, personal distress, distress about the child’s diabetes self-management, and health care team concerns. Cronbach’s alpha indicated adequate reliability. Higher PDDS-SP scores were associated with lower well-being and more frequent hypoglycemic episodes. Mothers reported higher distress than fathers. Conclusions: The PDDS-SP was found to be a reliable and valid measure for assessing diabetes-related distress among Spanish-speaking parents.

Jianan He, Joshua D. Butson, Ruijia Gu et al.

Abstract MXenes, a novel member of the 2D material family, shows promising potential in stabilizing isolated atoms and maximizing the atom utilization efficiency for catalytic applications. This review focuses on the role of MXenes as support for single‐atom catalysts (SACs) for various electrochemical reactions, namely the hydrogen evolution reaction (HER), oxygen evolution reaction (OER), oxygen reduction reaction (ORR), carbon dioxide reduction reaction (CO2RR), and nitrogen reduction reaction (NRR). First, state‐of‐the‐art characterization and synthesis methods of MXenes and MXene‐supported SACs are discussed, highlighting how the unique structure and tunable functional groups enhance the catalytic performance of pristine MXenes and contribute to stabilizing SAs. Then, recent studies of MXene‐supported SACs in different electrocatalytic areas are examined, including experimental and theoretical studies. Finally, this review discusses the challenges and outlook of the utilization of MXene‐supported SACs in the field of electrocatalysis.

Guliashki Vassil, Kirilov Leoneed, Marinova Galia

The Job Shop Scheduling Problem (JSSP) attracts many researchers due to its combinatorial nature and its discovery in numerous practical applications. This type of problem is characterized by high computational complexity; therefore, solving large-sized problems is not accessible with exact optimization methods. Very often, real JSSP problems can be presented as Flexible Job Shop Scheduling Problems (FJSSP). For these problems, there are single-criterion and multi-criteria mathematical models. On the other hand, the ways to solve this type of problems include exact methods and heuristic or metaheuristic algorithms. This paper the aim to review the progress of research in the field of solving FJSSP over the last 10 years, as well as to show current trends for future scientific developments in this area.

Richard Katuramu, Joseph KB Matovu, Joseph Kirabira et al.

Abstract Purpose Despite the availability of antiretroviral therapy (ART), nearly a quarter of people living with HIV (PLHIV) die within six months after hospital discharge due to complications from AIDS-related illnesses. Timely linkage to ART clinics after hospital discharge is crucial in reducing this mortality. We performed a systematic review to collate the available evidence on the strategies used to improve linkage to ART clinics after hospital discharge to inform future interventions. Methods and materials We systematically searched PUBMED, web of science, google scholar, embase and cochrane central for randomized controlled trials and quasi-experimental intervention trials conducted from January 2006 to December 2024, involving PLHIV aged 18 years and above in Low and Middle Income Countries (LMICs). Studies were included if they: (i) collected data in or after 2006, (ii) used randomized controlled trials (RCTs) or quasi-experimental prospective designs with a control group, (iii) reported at least one of several potential outcomes related to linkage to HIV ART clinics, and (iv) reported at least one strategy used to link PLHIV to HIV care after hospital discharge. Risk of bias was assessed using the Cochrane “risk of bias” tool for RCTs and the ROBINS-I tool for non-randomized studies of interventions. We used a narrative synthesis of articles to describe the different strategies used to enhance linkage to HIV ART clinics after hospital discharge. Results From the initial pool of 3003 articles, nine papers were independently reviewed and four (4) met the inclusion criteria. All the studies were conducted between 2011 and 2024 and comprised three RCTs and one quasi-experimental study. All the articles exhibited a low risk of bias. Strategies used to improve linkage to HIV care ART clinics included use of mobile phone appointment reminders, patient health education during hospitalization, multiple counseling sessions after hospital discharge, and the use of incentives such as food parcels. Conclusion Only few studies from LMICs have investigated strategies for linkage to ART clinics among PLHIV after hospital discharge. All the identified studies had more than one strategy applied. Further implementation research is recommended to explore context-specific strategies and strengthen the evidence base for improving linkage to ART clinics following hospital discharge. Review registration The protocol for this review was prespecified and published in PROSPERO (registration number (CRD42018110036).

Seongil Han, Haemin Jung, Paul D. Yoo

Abstract The Basel Accord emphasizes the necessity of employing internal data models to manage key credit risk components, including Probability of Default (PD), Loss Given Default (LGD), and Exposure At Default (EAD). Among these, internal datasets are critical for estimating PD, a fundamental measure of borrower creditworthiness. Nevertheless, practical application often faces challenges due to incomplete datasets, which can skew analyses and undermine the accuracy of credit scoring models. Traditional approaches to addressing missing data, such as sample deletion or mean imputation, are widely used; however, they often prove insufficient for accurate prediction. Consequently, imputation methods are typically favored over deletion, as they allow for the full utilization of available data. Recent advancements have introduced more sophisticated techniques, such as Generative Adversarial Imputation Networks (GAIN), which utilize a generative adversarial network to model data distributions and impute missing values with greater precision than conventional methods. Building on these developments, this study proposes a novel imputation approach, SMART (Structured Missingness Analysis and Reconstruction Technique) for credit scoring datasets. SMART consists of two primary stages: first, it normalizes and denoises the dataset using randomized Singular Value Decomposition (rSVD), followed by the implementation of GAIN to impute missing values. Experimental results demonstrate that SMART significantly outperforms existing state-of-the-art methods, particularly in high missing data contexts (20%, 50%, and 80%), with improvements in imputation accuracy of 7.04%, 6.34%, and 13.38%, respectively. In conclusion, SMART represents a substantial advancement in handling incomplete credit scoring datasets, leading to more precise PD estimation and enhancing the robustness of credit risk management models.

Jenny Baur

The Ishango Bone, a prehistoric artifact dated to approximately 20,000 years ago and discovered near the Semliki River in what is now the Democratic Republic of Congo, has intrigued researchers for the past 75 years. The artifact displays sixteen groups of notches arranged in three columns. While its function remains debated, this study suggests that the first two columns consist exclusively of all prime or odd numbers between 9 and 21, with the exception of 15, which appears only in the third column as two grouped pairs. Five groupings totaling 30 could be identified, and their arrangement may follow a consistent pattern. Additional numerical relationships between all three columns can be interpreted to support all four basic arithmetic operations. It is hypothesized that the notches may have served as reference marker to lay out their values for storytelling or teaching in the form of mathematical art. This study aims to broaden perspectives on the Ishango Bone and its traditional interpretation as a simple tallying device, and to encourage a re-evaluation of the mathematical capabilities of prehistoric humans.

Jin Kim, Byunghwee Lee, Taekho You et al.

The rise of multimodal generative AI is transforming the intersection of technology and art, offering deeper insights into large-scale artwork. Although its creative capabilities have been widely explored, its potential to represent artwork in latent spaces remains underexamined. We use cutting-edge generative AI, specifically Stable Diffusion, to analyze 500 years of Western paintings by extracting two types of latent information with the model: formal aspects (e.g., colors) and contextual aspects (e.g., subject). Our findings reveal that contextual information differentiates between artistic periods, styles, and individual artists more successfully than formal elements. Additionally, using contextual keywords extracted from paintings, we show how artistic expression evolves alongside societal changes. Our generative experiment, infusing prospective contexts into historical artworks, successfully reproduces the evolutionary trajectory of artworks, highlighting the significance of mutual interaction between society and art. This study demonstrates how multimodal AI expands traditional formal analysis by integrating temporal, cultural, and historical contexts.

Guosheng Hu

Due to its unique computing principles, quantum computing technology will profoundly change the spectacle of color art. Focusing on experimental exploration of color qubit representation, color channel processing, and color image generation via quantum computing, this article proposes a new technical path for color computing in quantum computing environment, by which digital color is represented, operated, and measured in quantum bits, and then restored for classical computers as computing results. This method has been proved practicable as an artistic technique of color qubit representation and quantum computing via programming experiments in Qiskit and IBM Q. By building a bridge between classical chromatics and quantum graphics, quantum computers can be used for information visualization, image processing, and more color computing tasks. Furthermore, quantum computing can be expected to facilitate new color theories and artistic concepts.

Ewelina Gajewska

This study aims to evaluate machine intelligence through artistic creativity by employing a modified version of the Turing Test inspired by Lady Lovelace. It investigates two hypotheses: whether human judges can reliably distinguish AI-generated artworks from human-created ones and whether AI-generated art achieves comparable aesthetic value to human-crafted works. The research contributes to understanding machine creativity and its implications for cognitive science and AI technology. Participants with educational backgrounds in cognitive and computer science play the role of interrogators and evaluated whether a set of paintings was AI-generated or human-created. Here, we utilise parallel-paired and viva voce versions of the Turing Test. Additionally, aesthetic evaluations are collected to compare the perceived quality of AI-generated images against human-created art. This dual-method approach allows us to examine human judgment under different testing conditions. We find that participants struggle to distinguish between AI-generated and human-created artworks reliably, performing no better than chance under certain conditions. Furthermore, AI-generated art is rated as aesthetically as human-crafted works. Our findings challenge traditional assumptions about human creativity and demonstrate that AI systems can generate outputs that resonate with human sensibilities while meeting the criteria of creative intelligence. This study advances the understanding of machine creativity by combining elements of the Turing and Lovelace Tests. Unlike prior studies focused on laypeople or artists, this research examines participants with domain expertise. It also provides a comparative analysis of two distinct testing methodologies (parallel-paired and viva voce) offering new insights into the evaluation of machine intelligence.

Victor Ngo, Rachel, Ramchurn et al.

This paper presents an evaluation of 18 children's in-the-wild experiences with the autonomous robot arm performer NED (Never-Ending Dancer) within the Thingamabobas installation, showcased across the UK. We detail NED's design, including costume, behaviour, and human interactions, all integral to the installation. Our observational analysis revealed three key challenges in child-robot interactions: 1) Initiating and maintaining engagement, 2) Lack of robot expressivity and reciprocity, and 3) Unmet expectations. Our findings show that children are naturally curious, and adept at interacting with a robotic art performer. However, our observations emphasise the critical need to optimise human-robot interaction (HRI) systems through careful consideration of audience's capabilities, perceptions, and expectations, within the performative arts context, to enable engaging and meaningful experiences, especially for young audiences.

E. Feigenbaum

Andrei Mihalea, Adina Magda Florea

Computer vision has significantly influenced recent advancements in autonomous driving by providing cutting-edge solutions for various challenges, including object detection, semantic segmentation, and comprehensive scene understanding. One specific challenge is ego-vehicle trajectory segmentation, which involves learning the vehicle’s path and describing it with a segmentation map. This can play an important role in both autonomous driving and advanced driver assistance systems, as it enhances the accuracy of perceiving and forecasting the vehicle’s movements across different driving scenarios. In this work, we propose a deep learning approach for ego-trajectory segmentation that leverages a state-of-the-art segmentation network augmented with guidance cues provided through various merging mechanisms. These mechanisms are designed to direct the vehicle’s path as intended, utilizing training data obtained with a self-supervised approach. Our results demonstrate the feasibility of using self-supervised labels for ego-trajectory segmentation and embedding directional intentions within the network’s decisions through image and guidance input concatenation, feature concatenation, or cross-attention between pixel features and various types of guidance cues. We also analyze the effectiveness of our approach in constraining the segmentation outputs and prove that our proposed improvements bring major boosts in the segmentation metrics, increasing IoU by more than 12% and 5% compared with our two baseline models. This work paves the way for further exploration into ego-trajectory segmentation methods aimed at better predicting the behavior of autonomous vehicles.

Václav Diviš, Bastian Spatz, Marek Hrúz

Recent research has drawn attention to the ambiguity surrounding the definition and learnability of Out-of-Distribution recognition. Although the original problem remains unsolved, the term “Out-of-Model Scope” detection offers a clearer perspective. The ability to detect Out-of-Model Scope inputs is particularly beneficial in safety-critical applications such as autonomous driving or medicine. By detecting Out-of-Model Scope situations, the system’s robustness is enhanced and it is prevented from operating in unknown and unsafe scenarios. In this paper, we propose a novel approach for Out-of-Model Scope detection that integrates three sources of information: (1) the original input, (2) its latent feature representation extracted by an encoder, and (3) a synthesized version of the input generated from its latent representation. We demonstrate the effectiveness of combining original and synthetically generated inputs to defend against adversarial attacks in the computer vision domain. Our method, TRust Your GENerator (TRYGEN), achieves results comparable to those of other state-of-the-art methods and allows any encoder to be integrated into our pipeline in a plug-and-train fashion. Through our experiments, we evaluate which combinations of the encoder’s features are most effective for discovering Out-of-Model Scope samples and highlight the importance of a compact feature space for training the generator.

Maria João Oliveira, Pedro Ribeiro, Pedro Miguel Rodrigues

Background: Alzheimer’s disease (AD) is a progressive and irreversible neurodegenerative condition that increasingly impairs cognitive functions and daily activities. Given the incurable nature of AD and its profound impact on the elderly, early diagnosis (at the mild cognitive impairment (MCI) stage) and intervention are crucial, focusing on delaying disease progression and improving patients’ quality of life. Methods: This work aimed to develop an automatic sMRI-based method to detect AD in three different stages, namely healthy controls (CN), mild cognitive impairment (MCI), and AD itself. For such a purpose, brain sMRI images from the ADNI database were pre-processed, and a set of 22 texture statistical features from the sMRI gray-level co-occurrence matrix (GLCM) were extracted from various slices within different anatomical planes. Different combinations of features and planes were used to feed classical machine learning (cML) algorithms to analyze their discrimination power between the groups. Results: The cML algorithms achieved the following classification <i>accuracy</i>: 85.2% for <i>AD</i> vs. <i>CN</i>, 98.5% for <i>AD</i> vs. <i>MCI</i>, 95.1% for <i>CN</i> vs. <i>MCI</i>, and 87.1% for <i>all</i> vs. <i>all</i>. Conclusions: For the pair <i>AD</i> vs. <i>MCI</i>, the proposed model outperformed state-of-the-art imaging source studies by 0.1% and non-imaging source studies by 4.6%. These results are particularly significant in the field of AD classification, opening the door to more efficient early diagnosis in real-world settings since MCI is considered a precursor to AD.