Domain Knowledge Research Articles

What Do We Mean When We Say "Insight"? A Formal Synthesis of Existing Theory.

Researchers have derived many theoretical models for specifying users' insights as they interact with a visualization system. These representations are essential for understanding the insight discovery process, such as when inferring user interaction patterns that lead to insight or assessing the rigor of reported insights. However, theoretical models can be difficult to apply to existing tools and user studies, often due to discrepancies in how insight and its constituent parts are defined. This article calls attention to the consistent structures that recur across the visualization literature and describes how they connect multiple theoretical representations of insight. We synthesize a unified formalism for insights using these structures, enabling a wider audience of researchers and developers to adopt the corresponding models. Through a series of theoretical case studies, we use our formalism to compare and contrast existing theories, revealing interesting research challenges in reasoning about a user's domain knowledge and leveraging synergistic approaches in data mining and data management research.

Free-listing and Semantic Knowledge: A Tool for Detecting Alzheimer Disease?

Impairment in semantic knowledge contributes to Alzheimer disease (AD)-related decline. However, the particulars of the impact AD has on specific domains of knowledge remain debatable. To investigate the impact of AD on specific semantic categories that are integral to daily functions-living things and man-made objects. We administered a free-listing task (written version) to 19 individuals with AD and 15 cognitively normal older adults and assessed the task's relationship with other cognitive and functional tests in clinical use. We compared the contents of the lists of salient concepts generated by the AD and control groups. Group membership (AD or control), after controlling for age, sex, formal education, and an estimate of premorbid intellectual ability, predicted the groups' performance on the free-listing task across two categories. Functional status was inversely related to performance on the free-listing task, holding demographic variables constant. Based on a comparison of the contents of the free lists that were generated by the two groups, it was possible to conclude that, in individuals with AD, conceptual knowledge central to the respective categories was well preserved, whereas the peripheral conceptual material showed evidence of degradation. The free-listing task, which is an easy-to-administer and cost-effective tool, could aid in the preliminary detection of semantic knowledge dysfunction, revealing concepts that are better preserved and, possibly, the characterization of AD. Cognitive assessment tools that can be applied across cultures are needed, and the free-listing task has the potential to address this gap.

Portfolio Optimization Strategies: New Approaches Based on Machine Learning Forecasting

This study provides an in-depth discussion and comprehensive review of the latest applications of machine learning techniques in the field of portfolio optimization. The article begins with an overview of traditional portfolio optimization theory and its limitations, and then focuses on how machine learning predictive models, which have flourished in recent years, can provide new perspectives and tools for solving the problems of non-linearity, dynamics and uncertainty in investment decision-making. This paper provides a detailed overview of the application practices of various machine learning algorithms (e.g., deep learning, reinforcement learning, integrated learning, etc.) in the areas of asset return prediction, risk assessment, and optimal weight allocation, and analyses their advantages and challenges compared to traditional methods. The analysis of relevant research cases reveals the significant effect of machine learning predictions in increasing expected portfolio returns, reducing risk exposure, and achieving effective diversification. The study also explores possible future trends and potential research directions for machine learning in portfolio optimization, highlighting the importance of combining domain knowledge with big data-driven intelligent investment decisions. This review aims to provide financial scholars and practitioners with a new way of thinking about portfolio optimization, and to promote the combination of theoretical research and practical operation in the fields of financial engineering and investment management, so as to achieve more accurate and efficient investment decisions.

A genetic programming approach with adaptive region detection to skin cancer image classification

Dermatologists typically require extensive experience to accurately classify skin cancer. In recent years, the development of computer vision and machine learning has provided new methods for assisted diagnosis. Existing skin cancer image classification methods have certain limitations, such as poor interpretability, the requirement of domain knowledge for feature extraction, and the neglect of lesion area information in skin images. This paper proposes a new genetic programming (GP) approach to automatically learn global and/or local features from skin images for classification. To achieve this, a new function set and a new terminal set have been developed. The proposed GP method can automatically and flexibly extract effective local/global features from different types of input images, thus providing a comprehensive description of skin images. A new region detection function has been developed to select the lesion areas from skin images for feature extraction. The performance of this approach is evaluated on three skin cancer image classification tasks, and compared with three GP methods and six non-GP methods. The experimental results show that the new approach achieves significantly better or similar performance in most cases. Further analysis validates the effectiveness of our parameter settings, visualizes the multiple region detection functions used in the individual evolved by the proposed approach, and demonstrates its good convergence ability.

Quantitatively determining melting properties for energetic compounds via knowledge-infused molecular graphs and interpretable deep learning

The melting properties of energetic compounds are critical to their performances, but challenges persist in understanding the molecular features and design strategies that drive these properties. Integrating domain knowledge into data-driven approaches for predicting melting properties enhances the generation of comprehensive insights and enables the construction of interpretable prediction models. For this purpose, a knowledge-infused molecular graphs (KIMGs) were devised to describe the characters of energetic compounds, by which prediction models were developed in conjunction with the message passing neural networks. A melting-point dataset composed of around 30,000 melt-castable energetic compounds was constructed, and a collection of 29 key descriptors relevant to melting behaviors is integrated in KIMGs. The model achieved the best mean absolute error of 10.93 K for the melting point prediction. The model interpretability from both molecular graphs and feature importances offer insights for understanding the complex interplay that determine the melting properties of energetic compounds. This work enables researchers not only to predict melting behaviors with enhanced accuracy but also applicable to establishing other quantitively structure-property relationships.

Grounded Relational Inference: Domain Knowledge Driven Explainable Autonomous Driving

Grounded Relational Inference: Domain Knowledge Driven Explainable Autonomous Driving

A Bibliometric Exploration of Multiple Criteria Decision Aid and Clustering—A Conceptual Taxonomy

ABSTRACTThis work explores the intersection of Multiple Criteria Decision Aid (MCDA) and clustering techniques, revealing unexploited potential and novel perspectives arising from their integration, challenging their conventional separation. It serves as a compass, guiding researchers through a bibliometric exploration and a conceptual taxonomy consolidating existing knowledge. Employing a two‐fold methodology, we first sketch the field's contours through a bibliometric lens, uncovering its intellectual structure, thematic landscape, and social dynamics. Then, using science mapping techniques like co‐word analysis, historiography, and collaboration network analysis, we examine patterns, revealing an interconnected mosaic of concepts. Our findings unveil a natural grouping into three categories: (1) Mixed‐yet‐not‐integrated approaches, explores sequential applications—clustering followed by MCDA or vice versa—where one method precedes and informs the other. (2) ‘Relational/ordered clustering’ leveraging criteria dependency to refine structures. (3) Using MCDA to improve clustering mechanics through similarity metrics, domain knowledge incorporation, and robustness. We conclusively propose a taxonomy along three axes: Units of Analysis, Instrumentalisation, and Objective. The key takeaway emphasises the collaborative potential of MCDA, envisioning a landscape where the integration of MCDA and clustering not only enhances existing methodologies but also spawns innovative paradigms, fostering a symbiotic relationship that transcends conventional boundaries.

Threat intelligence named entity recognition techniques based on few-shot learning

In today’s digital and internet era, threat intelligence analysis is of paramount importance to ensure network and information security. Named Entity Recognition (NER) is a fundamental task in natural language processing, aimed at identifying and extracting specific types of named entities from text, such as person names, locations, organization names, dates, times, currencies, and more. The quality of entities determines the effectiveness of upper-layer applications such as knowledge graphs. Recently, there has been a scarcity of training data in the threat intelligence field, and single models suffer from poor generalization ability. To address this, we propose a multi-view learning model, named the Few-shot Threat Intelligence Named Entity Recognition Model (FTM). We enhance the fusion method based on FTM, and further propose the FTM-GRU (Gate Recurrent Unit) model. The FTM model is based on the Tri-training algorithm to collaboratively train three few-shot NER models, leveraging the complementary nature of different model views to enable them to capture more threat intelligence domain knowledge at the coding level.FTM-GRU improves the fusion of multiple views. FTM-GRU uses the improved GRU model structure to control the memory and forgetting of view information, and introduces a relevance calculation unit to avoid redundancy of view information while highlighting important semantic features. We label and construct a few-shot Threat Intelligence Dataset (TID), and experiments on TID as well as the publicly available National Vulnerability Database (NVD) validate the effectiveness of our model for NER in the threat intelligence domain. Experimental results demonstrate that our proposed model achieves better recognition results in the task.

A Novel Hybrid Deep Learning Framework for Evaluating Field Evapotranspiration Considering the Impact of Soil Salinity

AbstractAccurate evaluation of evapotranspiration (ET) is crucial for efficient agricultural water management. Data‐driven models exhibit strong predictive ET capabilities, yet significant limitations like naive extrapolation hamper wider generalization. In this perspective, we explore a novel hybrid deep learning (DL) framework to integrate domain knowledge and demonstrate its potential for evaluating ET under the influence of soil salinity. Specifically, we integrated physical constraints from process models (Penman‐Monteith or Shuttleworth‐Wallace) and salinity‐induced stomatal stress mechanisms into the DL algorithm, and evaluated its performance by comparing four diverse scenarios. Results demonstrate that hybrid DL framework offers a promising alternative for ET estimation, achieving comparable accuracy to pure DL during training and validation. Nonetheless, due to the limited available measurements, data‐driven model may not adequately capture plant responses to salt stress, leading to significant prediction biases observed during independent testing. Encouragingly, the hybrid DL model (DL‐SS) integrating Shuttleworth‐Wallace and salinity‐induced stomatal stress mechanisms demonstrated enhanced interpretability, generalizability, and extrapolation capabilities. During testing, DL‐SS consistently showed optimal performance, yielding root mean square error (RMSE) values of 37.4 W m−2 for sunflower and 39.2 W m−2 for maize. Compared to traditional Jarvis‐type approaches (JPM and JSW) and pure DL model during testing, DL‐SS achieved substantial reductions in RMSE values: 51%, 33%, and 43% for sunflower, and 45%, 31%, and 35% for maize, respectively. These findings highlight the importance of integrating prior scientific knowledge into data‐driven models to enhance extrapolation capability of ET modeling, especially in salinized regions where conventional models may struggle.

Contrastive meta-reinforcement learning for heterogeneous graph neural architecture search

Heterogeneous Graph Neural Networks (HGNNs) have demonstrated significant success in capturing complex interactions within heterogeneous graphs to learn graph representations. However, designing effective HGNN architecture is a challenging task that demands substantial domain knowledge and human effort. Fortunately, the advent of Heterogeneous Graph Architecture Search (HGNAS) has automated this process and yielded networks surpassing those designed by humans. Nonetheless, conventional HGNAS approaches are limited in their ability to generalize to unseen or diverse task types. To address these limitations, we propose an efficient method called Contrastive Meta-reinforcement learning-based Heterogeneous Graph Neural Architecture Search (CM-HGNAS). Our approach aims to produce high-performance networks tailored to different types of tasks. Firstly, we use gradient-based meta-learning to rapidly adapt to new tasks by leveraging the knowledge acquired from meta-training tasks. Secondly, since our meta-training tasks encompass various types of tasks, including node classification and link prediction, we employ contrastive learning to unify the evaluation metric across all tasks. This unification enhances the generalization capabilities of our model across diverse tasks. Experimental results demonstrate that our method successfully generalizes to unseen tasks and surpasses existing HGNN baselines and HGNAS techniques in terms of performance.

The post-non-classical university’s identity: an axiological aspect

Introduction. The relevance of the university’s role reconsideration is a consequence of the classical university foundations “ruinization” in the contemporary functioning conditions, characterized as highly dynamic, fluid and complex. The identity of the post-non-classical university is not perceived as something integral or self-evident, it is represented the essence of multifaceted identity, lying in the opposition of the idea of university modeling to the destructive postmodern conditions in which the contemporary university transforms. The aim of the study is to conceptualize the post-non-classical university’s image in its axiological foundations Research methods. To reach the aim of the study, the authors applied a methodology comprising philosophically and culturologically determined methods of qualitative analysis to explore issues in the relevant knowledge domains and the direct praxeological aspect of visible manifestations of the modern university’s identity. In their argumentation, the authors rely on philosophical and cultural comparative analysis, use the techniques of systemic functional and axiological approaches, the method of conceptualizing the university’s image. Research results. The article considers the transition from the classical to the non-classical and post-non-classical model of the university. The reflexive (convergent) consciousness is regarded as an essential condition the identity of contemporary university constructing crisis overcoming, and the glocalization argued as the strategy of the university as a sociocultural institute. The post-non-classical university multifaceted identity integrates cultural, behavioral and representational aspects, remaining in its globalized structure the regional particularities. The graduate, possessing the possibility of self-image constructing to achieve integrity and additivity, becomes the axiological foundation of the post-non-classical identity. Conclusion. The study has shown that the contemporary university independently determines the reflexive modernization of the axiological foundations of self-identity. It does not react to external circumstances and challenges, but rather absorbs external conditions and circumstances, and actively transforms and pragmatically uses them for its own advancement. The post-non-classical identity of the university, on the one hand, is a product of postmodernity. On the other hand, its contemporary reflexive modeling overcomes the deconstructing and ruining foundations of postmodernity, and can be considered as a foresight technology

Relationship Between Original Research Experiences and Evidence-Based Practice Among Undergraduate Dental Hygiene Students.

For the dental hygiene profession, a need exists to engage students in research experiences as undergraduate students to foster interest in research and improve the implementation of evidence-based practice. The purpose was to explore the relationships between original research experiences and evidence-based practice among undergraduate dental hygiene students. This quantitative survey research study of undergraduate dental hygiene students in the USA was conducted in 2021. Original research experience was measured using the Wessex Research Network (WReN) spider instrument and evidence-based practice was measured using the Knowledge, Attitudes, Access and Confidence Evaluation (KACE) instrument. Invitation e-mails were sent to the directors of US entry-level dental hygiene programs to forward the survey e-mail message to students in the last semester of their dental hygiene programs. The data were analysed using descriptive statistics, bivariate analysis and multiple linear regression analysis. A total of 128 responses were analysed. The foundational knowledge domain of research experience was found to be a significant predictor of evidence-based practice and the mediators of evidence-based knowledge, evidence-based attitudes, evidence-based access and evidence-based confidence. Research experience was found to be a significant predictor of Evidence-Based Practice. Engaging undergraduate dental hygiene students in research opportunities may prepare students for the clinical application of evidence-based practice and original research projects.

AI Governance in Healthcare: Best Practices, Solutions, and Unresolved Issues

What challenges are presented by AI adoption in healthcare? What role should AI governance play in regard to the opportunities and risks for generative AI (large language models like ChatGPT-4, etc.) in health? How do we responsibly evaluate new technology and ensure guardrails for its deployment, but without stifling innovation? How do we address societal fears and biases? What level of reliability and robustness should algorithms achieve before being adopted with confidence across healthcare? How can AI support health equity and expand access? We will explore these questions and more with some of the nation’s leading experts on these topics. Learning Objectives Understand key opportunities and challenges for AI governance in healthcare, particularly in regard to rapidly emerging generative AI applications Discuss various responsible AI and AI governance frameworks that seek to ensure safety and efficacy in AI adoption Explore the role of government, legislation, and regulators in supporting safe and equitable adoption of AI in healthcare Explore opportunities for government, industry, and academic partnerships to accelerate AI validation, development, and adoption of safety standards Consider the potential impact of AI on the healthcare workforce and provider and patient experiences Identify the risks we should be most concerned with and what approaches can we take to mitigate those risks? By the end of the presentation, participants should have a sense of the opportunities and challenges of adopting AI in healthcare, key risks and frameworks for mitigating them, and the potential role of government, academia, and industry in addressing these issues Generative AI application examples Patient Communication and Engagement Intelligent chatbots interacting with patients, answering their queries, providing health advice, and helping them manage their conditions Mental Health Support Provide psychological support to patients, helping them manage stress, anxiety, depression, and other mental health issues. Provide interventions based on CBT and other therapeutic techniques Automated Medical Documentation Note-taking can allowing doctors to focus more on patient care Education and Training Create educational content and personalized training for medical students and healthcare professionals, potentially simulating things like patient cases and providing guidance on diagnosis, etc. Synthesis and Summarization of Medical Research Make it easier for people to keep up with the latest developments Public Health Communication In a public health emergency, help craft health messages, FAQs, and timely and accurate updates for the public Generative AI challenges and limitations Ensuring the accuracy of AI-generated information Maintaining patient privacy Handling the ethical implications of AI and human-machine interactions in healthcare “Hallucinations” (false or inaccurate information) can occur when model has limited knowledge of a particular domain “Distraction” (losing track of earlier inputs) can occur when inputs go beyond allowable token count for model “context window” and models start to produce responses not relevant to queries and context.

Information based explanation methods for deep learning agents—with applications on large open-source chess models

With large chess-playing neural network models like AlphaZero contesting the state of the art within the world of computerised chess, two challenges present themselves: the question of how to explain the domain knowledge internalised by such models, and the problem that such models are not made openly available. This work presents the re-implementation of the concept detection methodology applied to AlphaZero, by using large, open-source chess models with comparable performance. We obtain results similar to those achieved when applying this methodology to AlphaZero, while relying solely on open-source resources. We also present a novel explainable AI (XAI) method, which is guaranteed to highlight exhaustively and exclusively the information used by the explained model. This method generates visual explanations tailored to domains characterised by discrete input spaces, as is the case for chess. Our presented method has the desirable property of controlling the information flow between any input vector and the given model, which in turn provides strict guarantees regarding what information is used by the trained model during inference. We demonstrate the viability of our method by applying it to standard 8×8\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$8 \ imes 8$$\\end{document} chess, using large open-source chess models.

A3Test: Assertion-Augmented Automated Test case generation

Context:Test case generation is a critical yet challenging task in software development. Recently, AthenaTest – a Deep Learning (DL) approach for generating unit test cases has been proposed. However, our revisiting study reveals that AthenaTest can generate less than one-fifth of the test cases correctly, due to a lack of assertion knowledge and test signature verification. Objective:This paper introduces A3Test, a novel DL-based approach to the generation of test cases, enhanced with assertion knowledge and a mechanism to verify consistency of the name and signatures of the tests. A3Test aims to adapt domain knowledge from assertion generation to test case generation. Method:A3Test employs domain adaptation principles and introduces a verification approach to name consistency and test signatures. We evaluate its effectiveness using 5,278 focal methods from the Defects4j dataset. Results:Our findings indicate that A3Test outperforms AthenaTest and ChatUniTest. A3Test generates 2.16% to 395.43% more correct test cases, achieves 2.17% to 34.29% higher method coverage, and 25.64% higher line coverage. A3Test achieves 2.13% to 12.20% higher branch coverage, 2.22% to 12.20% higher mutation scores, and 2.44% to 55.56% more correct assertions compared to both ChatUniTest and AthenaTest respectively for one iteration. When generating multiple test cases per method A3Test still shows improvements and comparable efficacy to ChatUnitTest. A survey of developers reveals that the majority of the participants 70.51% agree that test cases generated by A3Test are more readable than those generated by EvoSuite. Conclusions:A3Test significantly enhances test case generation through its incorporation of assertion knowledge and test signature verification, contributing to the generation of correct test cases.

Interpretable Representation and Customizable Retrieval of Traffic Congestion Patterns Using Causal Graph-Based Feature Associations

The substantial increase in traffic data offers new opportunities to inspect traffic congestion dynamics from different perspectives. This paper presents a novel framework for the interpretable representation and customizable retrieval of traffic congestion patterns using causal relation graphs, which harnesses many of these opportunities. By integrating domain knowledge with innovative data management techniques, we address the challenges of effectively handling and retrieving the growing volume of traffic data for diverse analytical purposes. The framework leverages causal graphs to encode traffic congestion patterns, capturing fundamental phenomena and their spatiotemporal relationships, thus facilitating an interpretable high-level view of traffic dynamics. Moreover, a customizable similarity measurement function is introduced based on inexact graph matching, allowing users to tailor the retrieval process to specific requirements. This framework’s capability to retrieve customizable and interpretable congestion patterns is demonstrated through extensive experiments with real-world highway traffic data in the Netherlands, highlighting its value in supporting diverse data-driven studies and applications.

Causal effects of policy and occupant behavior on cooling energy

Rising energy consumption in buildings has prompted architectural engineers and policymakers to ask “what-if” questions in the pursuit of energy efficiency. Current simulation and machine learning methods rely primarily on correlations and associations necessitating causal inference techniques. In this research, a causal inferential approach combining double machine learning and domain knowledge using directed acyclic graphs is employed. The utility of this approach is demonstrated on the 2015 United States Residential energy consumption survey to evaluate the impact of energy policies and occupant behavior on cooling energy consumption. The results show that energy policies such as energy audits, proper insulation, access to interval data and Energy Star qualified windows significantly reduce energy use intensity (EUI). Additionally, air conditioning usage by occupants causally affects EUI. For instance, proper insulation reduces EUI by 5.603 kWh/m2 while changing static thermostat settings to automatic adjustments at certain times reduces EUI by 3.542 kWh/m2. This study suggests potential energy-saving policies for balanced building energy efficiency and functionality.

Analisis Morfologi Fi’il Madhi Dan Fi’il Mudhori Pada Ayat Al-Qur’an Surat An-Naba Juz 30

Written, spoken, and arbitrary, Arabic has a distinct language style. In the same way, there are many domains for Arabic knowledge. You'll be exposed to a wide range of subjects when studying Arabic. Morphology, often known as shorof science, is one genre. In terms of morphology, the goal of this study is to characterise the linguistic features of surah An-Naba' in the Qur'an. Library research is the method used in this kind of study. This research seeks to determine the number of fi'il mudhori in Surah An-naba juz 30 and their reading pattern, in addition to identifying the dhomir that appears in each fi'il mudhori sentence. And describes in terms of the nahwu and shorof science approaches. The research results illustrate that researchers found 16 words that included fiil mudhari, these 16 words had been included in several different wazans and dhomir-dhomir.

Myo-regressor Deep Informed Neural NetwOrk (Myo-DINO) for fast MR parameters mapping in neuromuscular disorders

Magnetic Resonance (MR) parameters mapping in muscle Magnetic Resonance Imaging (mMRI) is predominantly performed using pattern recognition-based algorithms, which are characterised by high computational costs and scalability issues in the context of multi-parametric mapping.Deep Learning (DL) has been demonstrated to be a robust and efficient method for rapid MR parameters mapping. However, its application in mMRI domain to investigate Neuromuscular Disorders (NMDs) has not yet been explored. In addition, data-driven DL models suffered in interpretation and explainability of the learning process. We developed a Physics Informed Neural Network called Myo-Regressor Deep Informed Neural NetwOrk (Myo-DINO) for efficient and explainable Fat Fraction (FF), water-T2 (wT2) and B1 mapping from a cohort of NMDs.A total of 2165 slices (232 subjects) from Multi-Echo Spin Echo (MESE) images were selected as the input dataset for which FF, wT2,B1 ground truth maps were computed using the MyoQMRI toolbox. This toolbox exploits the Extended Phase Graph (EPG) theory with a two-component model (water and fat signal) and slice profile to simulate the signal evolution in the MESE framework. A customized U-Net architecture was implemented as the Myo-DINO architecture. The squared L2 norm loss was complemented by two distinct physics models to define two ‘Physics-Informed’ loss functions: Cycling Loss 1 embedded a mono-exponential model to describe the relaxation of water protons, while Cycling Loss 2 incorporated the EPG theory with slice profile to model the magnetization dephasing under the effect of gradients and RF pulses. The Myo-DINO was trained with the hyperparameter value of the 'Physics-Informed' component held constant, i.e. λmodel = 1, while different hyperparameter values (λcnn) were applied to the squared L2 norm component in both the cycling loss. In particular, hard (λcnn=10), normal (λcnn=1) and self-supervised (λcnn=0) constraints were applied to gradually decrease the impact of the squared L2 norm component on the ‘Physics Informed’ term during the Myo-DINO training process.Myo-DINO achieved higher performance with Cycling Loss 2 for FF, wT2 and B1 prediction. In particular, high reconstruction similarity and quality (Structural Similarity Index > 0.92, Peak Signal to Noise ratio > 30.0 db) and small reconstruction error (Normalized Root Mean Squared Error < 0.038) to the reference maps were shown with self-supervised weighting of the Cycling Loss 2. In addition muscle-wise FF, wT2 and B1 predicted values showed good agreement with the reference values. The Myo-DINO has been demonstrated to be a robust and efficient workflow for MR parameters mapping in the context of mMRI. This provides preliminary evidence that it can be an effective alternative to the reference post-processing algorithm. In addition, our results demonstrate that Cycling Loss 2, which incorporates the Extended Phase Graph (EPG) model, provides the most robust and relevant physical constraints for Myo-DINO in this multi-parameter regression task. The use of Cycling Loss 2 with self-supervised constraint improved the explainability of the learning process because the network acquired domain knowledge solely in accordance with the assumptions of the EPG model.

Enhancing wind field resolution in complex terrain through a knowledge-driven machine learning approach

Wind energy, an essential component in the fight against climate change, relies heavily on precise, detailed wind field simulations to optimize wind farms, particularly in complex terrain with intricate wind patterns. However, conventional high-resolution simulations come with a hefty computational cost, limiting their applicability in real-time decision-making. This research addresses this challenge by proposing a machine learning-driven, computationally efficient approach utilizing a modified Generative Adversarial Network. The contribution of this work is threefold. Firstly, we provide access to a unique high-resolution dataset of wind fields in complex terrain. Secondly, we introduce a knowledge-based modification to the loss function, ensuring that the algorithm captures crucial characteristics of the flow within complex terrains. Finally, we demonstrate the potential of our approach to enhance wind flow resolution in real-life wind farms. Through this, our method delivers comparable accuracy to high-resolution simulations while substantially reducing computational demands. This advancement greatly enhances the accessibility and efficiency of high-resolution wind field simulations, facilitating real-time optimization of wind farms. Moreover, we illustrate that by designing an appropriate loss function informed by domain knowledge, we can mitigate the need for adversarial training.