Hasil untuk "Fine Arts"

Luan Vinícius Fiorio, Ivana Nikoloska, Bruno Defraene et al.

Sound source tracking is commonly performed using classical array-processing algorithms, while machine-learning approaches typically rely on precise source position labels that are expensive or impractical to obtain. This paper introduces a physics-guided variational model capable of fully unsupervised single-source sound source tracking. The method combines a variational encoder with a physics-based decoder that injects geometric constraints into the latent space through analytically derived pairwise time-delay likelihoods. Without requiring ground-truth labels, the model learns to estimate source directions directly from microphone array signals. Experiments on real-world data demonstrate that the proposed approach outperforms traditional baselines and achieves accuracy and computational complexity comparable to state-of-the-art supervised models. We further show that the method generalizes well to mismatched array geometries and exhibits strong robustness to corrupted microphone position metadata. Finally, we outline a natural extension of the approach to multi-source tracking and present the theoretical modifications required to support it.

Mansi, Avinash Kori, Francesca Toni et al.

Text guided diffusion models are used by millions of users, but can be easily exploited to produce harmful content. Concept unlearning methods aim at reducing the models' likelihood of generating harmful content. Traditionally, this has been tackled at an individual concept level, with only a handful of recent works considering more realistic concept combinations. However, state of the art methods depend on full finetuning, which is computationally expensive. Concept localisation methods can facilitate selective finetuning, but existing techniques are static, resulting in suboptimal utility. In order to tackle these challenges, we propose TRUST (Targeted Robust Selective fine Tuning), a novel approach for dynamically estimating target concept neurons and unlearning them through selective finetuning, empowered by a Hessian based regularization. We show experimentally, against a number of SOTA baselines, that TRUST is robust against adversarial prompts, preserves generation quality to a significant degree, and is also significantly faster than the SOTA. Our method achieves unlearning of not only individual concepts but also combinations of concepts and conditional concepts, without any specific regularization.

Falk Dippel, Yinan Yu, Annika Rosengren et al.

Transformers have defined the state-of-the-art for clinical prediction tasks involving electronic health records (EHRs). The recently introduced Mamba architecture outperformed an advanced Transformer (Transformer++) based on Llama in handling long context lengths, while using fewer model parameters. Despite the impressive performance of these architectures, a systematic approach to empirically analyze model performance and efficiency under various settings is not well established in the medical domain. The performances of six sequence models were investigated across three architecture classes (Transformers, Transformers++, Mambas) in a large Swedish heart failure (HF) cohort (N = 42820), providing a clinically relevant case study. Patient data included diagnoses, vital signs, laboratories, medications and procedures extracted from in-hospital EHRs. The models were evaluated on three one-year prediction tasks: clinical instability (a readmission phenotype) after initial HF hospitalization, mortality after initial HF hospitalization and mortality after latest hospitalization. Ablations account for modifications of the EHR-based input patient sequence, architectural model configurations, and temporal preprocessing techniques for data collection. Llama achieves the highest predictive discrimination, best calibration, and showed robustness across all tasks, followed by Mambas. Both architectures demonstrate efficient representation learning, with tiny configurations surpassing other large-scaled Transformers. At equal model size, Llama and Mambas achieve superior performance using 25% less training data. This paper presents a first ablation study with systematic design choices for input tokenization, model configuration and temporal data preprocessing. Future model development in clinical prediction tasks using EHRs could build upon this study's recommendation as a starting point.

Noshitha Padma Pratyusha Juttu, Sahithi Singireddy, Sravani Gona et al.

Large Language Models (LLMs) have transformed text understanding, yet their adaptation to specialized legal domains remains constrained by the cost of full fine-tuning. This study provides a systematic evaluation of fine tuning, parameter efficient adaptation (LoRA, QLoRA), and zero-shot prompting strategies for unfair clause detection in Terms of Service (ToS) documents, a key application in legal NLP. We finetune BERT and DistilBERT, apply 4-bit Low-Rank Adaptation (LoRA) to models such as TinyLlama, LLaMA 3B/7B, and SaulLM, and evaluate GPT-4o and O-versions in zero-shot settings. Experiments on the CLAUDETTE-ToS benchmark and the Multilingual Scraper Corpus show that full fine-tuning achieves the strongest precision recall balance, while LoRA-based models provide competitive recall with up to 3x lower memory cost. These findings highlight practical design trade-offs for efficient and domain-adapted LLMs, contributing open baselines for fine-tuning research in legal text processing.

Han Sheng, Xu Li, Shucai Zeng

An important element of urban landscapes is various plants, and contact with urban landscapes can promote children’s positive mood and mental health. However, few studies focus on Asian school-aged children’s mood for different urban landscapes and the factors shaping them. This study attempted to understand the variables, including plant color, familiarity, and viewing distances (setting 0 m and 2 m), using 150 landscape scenes (68 flowering plants, 50 exotic plants, and 32 foliage plants), on the effects of the landscape preferences and mood states of 119 school-aged children (55 boys and 64 girls). Then, using partial least squares path modelling analysis to display the gender difference in children’s color perception, landscape preferences, and mood states. The results show that: (1) Plant color richness, familiarity, and the proportion of non-green parts of scenes positively affected children’s mood states. (2) Flowering plants are more likely to produce positive moods than those of exotic plants and foliage plants. (3) Plant color richness and familiarity significantly and positively correlated with children’s mood states and landscape preferences. (4) Notably, gender differences exist in children’s landscape preferences and mood states. This study underscores the importance of plant color collocation in child-friendly landscapes and considers the gender differences in urban landscape policy decisions. Besides, adding flowering plants and native plants in urban landscapes may potentially enhance children’s mood state and urban green space utilization rate.

Claudia Pirina, Giovanni Comi, Vincenzo d’Abramo

The paper addresses the issue of energy transition through the regeneration and reuse of abandoned urban spaces. While the theme of re-naturalisation of soils and preservation of natural environments is a familiar one in contemporary architectural culture, the need to rethink the systematisation of individual interventions in order to arrive at a transcalar approach to design seems to be a reflection that still requires the necessary critical investigation. In this sense, the contribution, through the reading of some case studies, investigates the potential that elements such as urban voids and landscape-territorial systems, declined in their ability to affect the energy transition, are capable of generating for the project, contributing directly and indirectly on transformative phenomena, both in urban and peripheral areas.   Article info Received: 18/03/2024; Revised: 23/04/2024; Accepted: 07/05/2024

Sergey Troitskiy, Emil Babaev, Elizaveta Belova et al.

The study, conducted in March 2022, involved the analysis of the content in several social media chats and groups; the participants of those chats live in the same place and therefore have a common experience of the space. The study was based on the hypothesis of a direct connection between the mental map (a system of individual ideas about space), the cultural reputation of topoi, and urban trauma, embodied in the unease infrastructure. The problem of assessing the significance of a place was solved by means of folklore toponymies – the mechanism of renaming, which indicates the degree of awareness about a specific place and defines its location on the mental map as well as ascribes a certain status to it. These statuses demonstrate the degree of significance of a place for a certain subject and form a kind of hierarchy, a system of topographical preferences. Thanks to online communication, people can not only transmit information much faster than the traditional forms of folklore dissemination allow, but also broadcast personal attitudes, conveying them as a bundle of meanings (for example, while inventing new toponyms). Therefore, one of the objectives of the study was to identify established folklore toponyms in online communication: they serve as markers of attitudes, reputation, and significance; we also try to catalogue attempts to “rename” different places. Another task was to find the symptoms of such anxiety in online communication.

Ali Youssef, Kristina Vodorezova, Yannick Aarts et al.

IUMENTA (Latin for livestock) is an innovative software framework designed to construct and simulate digital twins of animals. By leveraging the powerful capability of the Open Digital Twin Platform (ODTP) alongside advanced software sensors, IUMENTA offers researchers a user-friendly tool to seamlessly develop adaptive digital replicas of animal-based processes. This framework establishes a dynamic ecosystem that integrates insights from diverse experiments, consequently enhancing our understanding of animal behavioural and physiological responses. Through real-time tracking of an animal's energy balance. IUMENTA provides valuable insights into metabolic rates, nutritional needs, emotional states, and overall well-being of animals. In this article, we explore the application of the IUMENTA framework in developing a digital twin focused on the animal's energy balance. IUMENTA includes the EnergyTag system, a state-of-the-art wearable software sensor, which facilitates real-time monitoring of energy expenditure, allowing for continuous updates and personalisation of the energy balance digital twin.

Yushi Yang, Andrew M. Bean, Robert McCraith et al.

Fine-tuning Large Language Models (LLMs) incurs considerable training costs, driving the need for data-efficient training with optimised data ordering. Human-inspired strategies offer a solution by organising data based on human learning practices. This study evaluates the fine-tuning efficiency of five human-inspired strategies across four language models, three datasets, and both human- and LLM-labelled data in the context of medical question answering. These strategies achieve the best accuracy gain of 1.81% and an average gain of 1.02% across datasets, with interleaved strategies delivering the best average results. However, the best strategy varies across model-dataset combinations, limiting the generalisability of the effects of any single strategy. Additionally, LLM-defined question difficulty outperforms human-defined labels in curriculum-based learning, showing the potential of model-generated data as a cost-effective alternative for optimising fine-tuning.

R. Rehm, J. S. G. Mombarg, C. Aerts et al.

Numerical computations of stellar oscillations for models representative of B-type stars predict fewer modes to be excited than observations reveal from modern space-based photometric data. One shortcoming of state-of-the-art evolution models of B-type stars that may cause a lack of excited modes is the absence of microscopic diffusion in most such models. We investigate whether the inclusion of microscopic diffusion in stellar models of B-type stars, notably radiative levitation experienced by isotopes, leads to extra mode driving by the opacity mechanism compared to the case of models that do not include microscopic diffusion. We consider the case of slowly to moderately rotating stars and use non-rotating equilibrium models, while we account for (uniform) rotation in the computations of the pulsation frequencies. We calculate 1D stellar models with and without microscopic diffusion and examine the effect of radiative levitation on mode excitation, for both low-radial order pressure and gravity modes and for high-radial order gravity modes. We find systematically more modes to be excited for the stellar models including microscopic diffusion compared to those without it, in agreement with observational findings of pulsating B-type dwarfs. Furthermore, the models with microscopic diffusion predict that excited modes occur earlier on in the evolution compared to modes without it. In order to maintain realistic surface abundances during the main sequence, we include macroscopic envelope mixing by internal gravity waves. While radiative levitation has so far largely been neglected in stellar evolution computations of B-type stars for computational convenience, it impacts mode excitation predictions for stellar models of such stars. We conclude that the process of radiative levitation is able to reduce the discrepancy between predicted and observed excited pulsation modes in B-type stars.

Marta Pérez Azcárate, Susana Duque Valero, Joan Ramon Aromi Folch et al.

The results of the conservation work carried out on an exhibition set-up dating from the early twentieth century are presented. The exhibition set-up consists of a wooden showcase containing about twenty evaporite rocks from the collection of the Museu de Ciències Naturals de Barcelona (Spain). The work involved the remedial conservation of the rock specimens and showcase, and the improvement of the original environmental control system using sustainability criteria. An interdisciplinary team worked on the different phases of the project, which included prior historical and environmental studies. The remedial conservation of all elements in the collection has improved its accessibility and the monitoring of the environmental conditions of the new installation has confirmed the efficiency of the proposed passive environmental control system.

Pohsun Wang, Wu Song, Junling Zhou et al.

Confronting the age of artificial intelligence, exploring art through technology has become one of the directions of interdisciplinary development. Not only does artificial intelligence technology explore sustainability on a technical level; it can also take advantage of itself to focus on the visual perception of the living environment. People frequently interpret environmental features through their eyes, and the use of intuitive eye-tracking can provide effective data that can contribute to environmental sustainability in managing the environment and color planning to enhance the image of cities. This research investigates the visual responses of people viewing the historic city of Macau through an eye movement experiment to understand how the color characteristics of the physical environment are perceived. The research reveals that the buildings and plantings in the historic district of Macau are the most visible objects in the environment, while the smaller scale of St. Dominic’s Square, the Company of Jesus Square, and St. Augustine’s Square, which have a sense of spatial extension, have also become iconic environmental landscapes. This also draws visual attention and guides the direction of travel. The overall impressions of the Historic Centre of Macau, as expressed by the participants after the eye movement experiment, were mainly described as “multiculturalism”, “architectural style”, “traditional architecture”, “color scheme”, and “garden planting”. The 60 colors representing the urban color of Macau are then organized around these deep feelings about the environment. Therefore, for future inspiration, the 60 colors can be applied through design practice to create color expressions that fit the local characteristics, and thereby enhance the overall visual image of the city.

Nadya Serebriakova, Andrew Tkachenko, Sarah Gebruers et al.

Modern stellar structure and evolution theory experiences a lack of observational calibrations for the interior physics of intermediate- and high-mass stars. This leads to discrepancies between theoretical predictions and observed phenomena mostly related to angular momentum and element transport. Analyses of large samples of massive stars connecting state-of-the-art spectroscopy to asteroseismology may provide clues on how to improve our understanding of their interior structure. We aim to deliver a sample of O- and B-type stars at metallicity regimes of the Milky Way and the Large Magellanic Cloud (LMC) galaxies with accurate atmospheric parameters from high-resolution spectroscopy, along with a detailed investigation of line-profile broadening, for future asteroseismic studies. After describing the general aims of our two Large Programs, we develop dedicated methodology to fit spectral lines and deduce accurate global stellar parameters from high-resolution multi-epoch UVES and FEROS spectroscopy. We use the best available atmosphere models for three regimes covered by our global sample, given its breadth in terms of mass, effective temperature, and evolutionary stage. Aside from accurate atmospheric parameters and locations in the Hertzsprung-Russell diagram, we deliver detailed analyses of macroturbulent line broadening, including estimation of the radial and tangential components. We find that these two components are difficult to disentangle from spectra with signal-to-noise ratios below 250. Future asteroseismic modelling of the deep interior physics of the most promising stars in our sample will improve the existing dearth of such knowledge for large samples of OB stars, including those of low metallicity in the LMC.

Adolfo F. L. Baratta

Victor S. Bursztyn, David Demeter, Doug Downey et al.

How to usefully encode compositional task structure has long been a core challenge in AI. Recent work in chain of thought prompting has shown that for very large neural language models (LMs), explicitly demonstrating the inferential steps involved in a target task may improve performance over end-to-end learning that focuses on the target task alone. However, chain of thought prompting has significant limitations due to its dependency on huge pretrained LMs. In this work, we present compositional fine-tuning (CFT): an approach based on explicitly decomposing a target task into component tasks, and then fine-tuning smaller LMs on a curriculum of such component tasks. We apply CFT to recommendation tasks in two domains, world travel and local dining, as well as a previously studied inferential task (sports understanding). We show that CFT outperforms end-to-end learning even with equal amounts of data, and gets consistently better as more component tasks are modeled via fine-tuning. Compared with chain of thought prompting, CFT performs at least as well using LMs only 7.4% of the size, and is moreover applicable to task domains for which data are not available during pretraining.

Diederik Aerts, Jonito Aerts Arguëlles

For two decades, the formalism of quantum mechanics has been successfully used to describe human decision processes, situations of heuristic reasoning, and the contextuality of concepts and their combinations. The phenomenon of 'categorical perception' has put us on track to find a possible deeper cause of the presence of this quantum structure in human cognition. Thus, we show that in an archetype of human perception consisting of the reconciliation of a bottom up stimulus with a top down cognitive expectation pattern, there arises the typical warping of categorical perception, where groups of stimuli clump together to form quanta, which move away from each other and lead to a discretization of a dimension. The individual concepts, which are these quanta, can be modeled by a quantum prototype theory with the square of the absolute value of a corresponding Schrödinger wave function as the fuzzy prototype structure, and the superposition of two such wave functions accounts for the interference pattern that occurs when these concepts are combined. Using a simple quantum measurement model, we analyze this archetype of human perception, provide an overview of the experimental evidence base for categorical perception with the phenomenon of warping leading to quantization, and illustrate our analyses with two examples worked out in detail.

Doaa Abouelmagd, Hager Mohamed Abdelrahman Ahmed

Many architecture schools do not apply experimental learning in teaching theoretical courses like the history of architecture courses. There is a gap between what the students learn in theoretical courses like architectural styles in history courses and design practices. History classes are taught with a teacher-oriented approach, where students miss active and experimental learning. As a result, in the Egyptian urban setting, we find buildings designed using architectural styles, mostly Classical designs, without respecting the technical, aesthetic, and human factors. Furthermore, nowadays, the Egyptian government is very active in constructing the new Administrative Capital of Egypt. The new city is located 45 kilometers East of Cairo, and it is planned to be Egypt's new business and financial center. The new Downtown of the new capital city, or "The new Garden City," is planned to reassemble Khedival Cairo's peculiar value buildings with the renaissance, neo-classic, eclectic, and neo-baroque styles. As architectural education cannot be separated from current architectural practices, this paper introduces the research project results in the course "History of Architecture-level two." The course was taught in the first semester of the academic year (2018-19) in Fine arts, Cairo, Helwan University, Egypt. The students were asked to select a heritage building with a peculiar value from Khedival Cairo, analyze it with the seven principles of design, and propose a design for a building's facades in the new Capital city Downtown.This paper aims to introduce the results of the students' work, analyzing their designs, and the effect of experimental learning in developing their intellectual skills. The paper includes the results of a designed questioner answered by the students to evaluate the research project and the link between the history of architecture, heritage, and architecture design. The paper shows the importance of applying experimental learning and linking theoretical and practical courses for achieving better architecture practice.

Loic Jezequel, Ngoc-Son Vu, Jean Beaudet et al.

Detecting anomalies using deep learning has become a major challenge over the last years, and is becoming increasingly promising in several fields. The introduction of self-supervised learning has greatly helped many methods including anomaly detection where simple geometric transformation recognition tasks are used. However these methods do not perform well on fine-grained problems since they lack finer features. By combining in a multi-task framework high-scale shape features oriented task with low-scale fine features oriented task, our method greatly improves fine-grained anomaly detection. It outperforms state-of-the-art with up to 31% relative error reduction measured with AUROC on various anomaly detection problems.

Justin Kay, Matt Merrifield

Camera-based electronic monitoring (EM) systems are increasingly being deployed onboard commercial fishing vessels to collect essential data for fisheries management and regulation. These systems generate large quantities of video data which must be reviewed on land by human experts. Computer vision can assist this process by automatically detecting and classifying fish species, however the lack of existing public data in this domain has hindered progress. To address this, we present the Fishnet Open Images Database, a large dataset of EM imagery for fish detection and fine-grained categorization onboard commercial fishing vessels. The dataset consists of 86,029 images containing 34 object classes, making it the largest and most diverse public dataset of fisheries EM imagery to-date. It includes many of the characteristic challenges of EM data: visual similarity between species, skewed class distributions, harsh weather conditions, and chaotic crew activity. We evaluate the performance of existing detection and classification algorithms and demonstrate that the dataset can serve as a challenging benchmark for development of computer vision algorithms in fisheries. The dataset is available at https://www.fishnet.ai/.

Qingyun Wang, Semih Yavuz, Victoria Lin et al.

Graph-to-text generation has benefited from pre-trained language models (PLMs) in achieving better performance than structured graph encoders. However, they fail to fully utilize the structure information of the input graph. In this paper, we aim to further improve the performance of the pre-trained language model by proposing a structured graph-to-text model with a two-step fine-tuning mechanism which first fine-tunes the model on Wikipedia before adapting to the graph-to-text generation. In addition to using the traditional token and position embeddings to encode the knowledge graph (KG), we propose a novel tree-level embedding method to capture the inter-dependency structures of the input graph. This new approach has significantly improved the performance of all text generation metrics for the English WebNLG 2017 dataset.