Speakers

Abstract: Instabilities A Hybrid CFD-Neural Network Framework for The Early Prediction of Thermo-Acoustic in Hydrogen-Fueled Gas Turbine Combustors

The transition to low-carbon energy systems has intensified interest in hydrogen as a carbon-free fuel for gas turbine power generation. However, hydrogen’s high flame speed, diffusivity, and reactivity significantly increase its susceptibility to thermo-acoustic instabilities (TAIs), which result from the coupling between unsteady heat release and acoustic pressure oscillations. These instabilities can cause excessive vibrations, structural damage, efficiency losses, and operational failure. Early detection remains challenging because high-fidelity Computational Fluid Dynamics (CFD) simulations are computationally expensive and unsuitable for real-time applications. This study presents a hybrid Computational Fluid Dynamics–Neural Network (CFD–NN) framework for early prediction of longitudinal thermo-acoustic instabilities in hydrogen-fueled combustors. A two-dimensional high-fidelity CFD model is developed to simulate unsteady hydrogen combustion. Time-resolved pressure and heat-release data are collected from six monitoring probes, and the Local Rayleigh Index (LRI) is used to identify instability-driving regions. Results show strong spatial variation, with one probe exhibiting dominant instability while others demonstrate damping or transitional behavior. CFD-derived features are used to train a feedforward neural network for early instability prediction. The model achieves excellent accuracy (R² = 0.9998) with significantly reduced computation time, delivering predictions hundreds of times faster than CFD. The proposed framework enables real-time monitoring and intelligent combustion control for hydrogen gas turbines.

Abstract: Estimating H I Mass Fraction in Galaxies with Bayesian Neural Networks

Neutral atomic hydrogen (H I) regulates galaxy growth and quenching, but direct 21 cm measurements remain observationally expensive and affected by selection biases. We develop Bayesian neural networks (BNNs)—a type of neural model that returns both a prediction and an associated uncertainty—to infer the H I mass, log10(MHI), from widely available optical properties (e.g., stellar mass, apparent magnitudes, and diagnostic colors) and simple structural parameters. For continuity with the photometric gas fraction (PGF) literature, we also report the gas-to-stellar-mass ratio, log10(G/S), where explicitly noted. Our dataset is a reproducible cross-match of SDSS DR12, the MPA–JHU value-added catalogs, and the 100% ALFALFA release, resulting in 31,501 galaxies after quality controls. To ensure fair evaluation, we adopt fixed train/validation/test partitions and an additional sky-holdout region to probe domain shift, i.e., how well the model extrapolates to sky regions that were not used for training. We also audit features to avoid information leakage and benchmark the BNNs against deterministic models, including a feed-forward neural network baseline and gradient-boosted trees (GBTs, a standard tree-based ensemble method in machine learning). Performance is assessed using mean absolute error (MAE), root-mean-square error (RMSE), and probabilistic diagnostics such as the negative log-likelihood (NLL, a loss that rewards models that assign high probability to the observed H I masses), reliability diagrams (plots comparing predicted probabilities to observed frequencies), and empirical 68%/95% coverage. The Bayesian models achieve point accuracy comparable to the deterministic baselines while additionally providing calibrated prediction intervals that adapt to stellar mass, surface density, and color. This enables galaxy-by-galaxy uncertainty estimation and prioritization for 21 cm follow-up that explicitly accounts for predicted uncertainties (“risk-aware” target selection). Overall, the results demonstrate that uncertainty-aware machine-learning methods offer a scalable and reproducible route to inferring galactic H I content from widely available optical data.

Abstract: Multi-agent Reinforcement Learning in Networking

Multi-agent systems (MAS) deal with systems where multiple agents must make decisions to
achieve their goals in a shared environment. Despite their ubiquity, their effective
implementation still faces several challenges. We will address these challenges by drawing on
recent advances in multi-agent reinforcement learning (MARL), that is, research on data
driven decision-making within MAS. We will also present some recent applications of MARL
in the context of energy-efficient resource allocation in networks.

Abstract: AI-Enhanced Digital Twins for Intelligent Manufacturing and Predictive Maintenance in Hydrogen Systems

Green hydrogen plays a central role in global decarbonization efforts, serving as a versatile energy carrier for transportation, industry, and energy storage. However, its widespread adoption is hindered by critical safety and reliability challenges, including high flamma- bility, hydrogen embrittlement, leak risks, and accelerated degradation of electrochemical components in systems such as proton exchange membrane (PEM) electrolyzers, storage tanks, pipelines, and refueling stations. This presentation introduces an integrated Digital Twin (DT) framework augmented by Artificial Intelligence (AI) to enable real-time mon- itoring, anomaly detection, predictive maintenance, and enhanced operational safety in hydrogen infrastructures. The approach combines physics-based modeling with advanced AI techniques, including machine learning, deep learning, and Physics-Informed Neural Networks (PINNs). A key focus is a laboratory-scale PEM electrolyzer (two-cell Titan EZ120 stack) mirrored through a DT architecture. The virtual replica, developed using Unity and Blender for interactive 3D visualization, synchronizes with experimental polar- ization data acquired via potentiostat under controlled temperatures (20–80 ◦C). PINNs are embedded within the DT, incorporating electrochemical constraints such as Butler- Volmer kinetics, Arrhenius temperature dependence, and ohmic resistance into a hybrid loss function. A feedforward neural network (4 hidden layers × 64 neurons, ReLU acti- vation) achieves a 12 % reduction in mean squared error compared to conventional neural networks, with RMSE values of 0.05–0.12 V and R² > 0.98 across conditions.  Real- time visualization occurs via a Grafana dashboard integrated with InfluxDB, supporting 10 Hz sampling and low end-to-end latency (150–250 ms). Broader industrial applications include predictive leak detection in refueling stations, degradation forecasting in elec- trolyzers (e.g., simulated overpotential increase of 15–20 % after 1000 cycles), structural health monitoring in pipelines and tanks, and risk assessment in underground storage using federated AI models. The framework addresses limitations such as data sparsity, model explainability, and cybersecurity through standardized DT architectures and open datasets. This work demonstrates how AI-driven DTs foster intelligent, resilient hydrogen ecosystems, accelerating safe deployment of green hydrogen technologies and contributing to sustainable energy transitions in alignment with Industry 4.0 principles.

Abstract: Trustworthy AI in Hospitals: A Unified Framework for Privacy, Robustness, and Human-Centered Explanations

Hospitals are rapidly integrating Artificial Intelligence (AI) into clinical documentation, decision support, and operational workflows. Yet, real-world adoption is increasingly constrained by a triad of concerns that are often addressed in isolation: (i) privacy-preserving learning and data minimization, (ii) security and robustness against adversarial manipulation, especially in Natural Language Processing (NLP) systems powered by Large Language Models (LLMs), and (iii) explanations that are meaningful to different human stakeholders in high-stakes environments. This abstract presents a unified framework for trustworthy hospital AI that jointly operationalizes privacy, robustness, and human-centered explanations, with a specific focus on LLM-enabled clinical assistants and multi-stakeholder use in routine care. First, we introduce a zero-trust prompt architecture for LLM-based clinical tasks such as summarization, triage support, and structured coding of unstructured notes. In this setting, the clinical text itself is treated as potentially adversarial input. The framework enforces strict separation between system policies and user-provided clinical content, employing structured prompting, context isolation, tool sandboxing, and provenance-aware retrieval to reduce susceptibility to prompt injection and instruction hijacking embedded in notes. To complement preventive controls, we include an adversarial evaluation protocol that stress-tests LLM behavior under realistic hospital threat scenarios (e.g., malicious or compromised documentation, data tampering, or accidental inclusion of unsafe instructions), reporting failure modes that are actionable for security and compliance teams. Second, we propose a role-adaptive explanation layer that generates differentiated explanations for clinicians, cybersecurity/IT stakeholders, and patients, recognizing that “one-size-fits-all” explainability can undermine trust. For clinicians, explanations emphasize clinically relevant evidence, uncertainty, and recommended next steps. For IT/security teams, explanations expose system integrity signals to support incident response and auditability. For patients, explanations prioritize clarity, limitations, and contextual guidance, aligned with human-centered communication principles. Across roles, explanations are designed to be faithful to model behavior while remaining usable under time pressure, and are accompanied by calibrated uncertainty cues and escalation pathways to human oversight

Abstract: Cultural Landscape Gene–Driven Multimodal 3D Perception and Knowledge-Graph Analytics for Cultural Heritage Landmarks: The Heluo Digital Intelligence Platform

Driven by globalization and rapid digital transformation, heritage conservation and utilization face increasing pressures, including architectural homogenization, uncontrolled urban–rural expansion that compresses cultural space, and the isolation of heritage sites in industrialized contexts. Meanwhile, within heritage tourism, the concept and value interpretation of “cultural landmarks” remain insufficiently articulated, and many digitalization efforts follow fragmented or single-track technical routes. Taking the Heluo region—an important cradle of Chinese civilization—as a case, this paper adopts the Cultural Landscape Gene (CLG) perspective to characterize cultural landmarks along three dimensions: historical accumulation, visual-form expression, and socio-cultural identity, and to explore a sustainable conservation–tourism synergy. We develop the “Heluo Digital Intelligence” platform, underpinned by a CLG management database and GIS-based spatial analytics, and implement a closed-loop workflow of collection, classification, evaluation, reconstruction, and presentation. Multi-source 3D digitization (UAV oblique photogrammetry, terrestrial/handheld 3D scanning, and SfM/MVS reconstruction with meshing and texturing) is integrated with AI to build gene profiles and to automatically extract features from images and 3D models. In addition, NLP-based named-entity recognition and knowledge graphs enable semantic interpretation and relational analysis of cultural genes, supporting digital exhibition, virtual restoration, and cultural IP development. A representative endangered-heritage case demonstrates the feasibility of long-term digital archiving and re-creation. The results indicate that the CLG perspective can bridge value interpretation, data governance, technical implementation, and decision support, offering a replicable pathway for heritage digitalization and sustainable use in comparable regions.

Abstract: Left Behind, Falling Behind, and Following: What Does the Older Generation Lose by Rejecting Artificial Intelligence?

Professional and academic discussions on artificial intelligence usually focus on younger generations, students, and digitally more open age groups, while far less attention is paid to older adults, including still-active workers and active retirees. Yet for them as well, an increasingly important question emerges: what professional, practical, and quality-of-life losses may result from the rejection of these technologies? The issue is also socially and economically relevant. In the context of adverse demographic trends and labour shortages, the experience and continued employment of older workers may become increasingly important. However, this can only be sustainable in the long term if they are able to keep pace with technological change and do not become detached from current developments. The presentation argues that distancing oneself from artificial intelligence in later life may mean not only digital disadvantage, but also narrower access to information, weaker autonomous decision-making, lower security awareness, and reduced professional adaptability. The presentation is based on qualitative research using in-depth interviews to explore older adults’ attitudes toward artificial intelligence. The background of the study lies partly in the fact that the author duo’s previous publications focused primarily on the relationship between Generation Z and AI. However, a new interview situation opened a different direction of inquiry. A highly educated vocational educator in his fifties asked one of the authors whether she had already used artificial intelligence in practice. This question led almost immediately to a spontaneous in-depth interview, during which the interviewee gradually recognized that AI could be especially useful in preparing persuasive and well- structured teaching materials. This experience inspired the broader research idea that, at a later stage of one’s career, the reflexive rejection of technology may become especially harmful. Rather than seeking statistical generalization, the presentation offers insight through illustrative interview excerpts and recurring patterns in respondents’ reflections. The exploratory analysis identifies three broad groups within the older generation: those who have already fallen behind, those who are at risk of falling behind, and those who have begun to follow technological change. The interviews suggest that AI can create value not only in work-related contexts but also in everyday life. Examples include interpreting simpler contractual situations in advance, which may save time and money in some cases; supporting household decisions, such as planning meals from ingredients already available at home; and finding podcasts aligned with personal interests and value preferences. AI also appeared as a tool for increasing security awareness: one interviewee asked it what kinds of telephone scams she should be especially careful about. An important ethical conclusion of the study is that AI should primarily be entrusted with tasks that users could also perform themselves, only more slowly, and in areas where they are capable of recognizing possible errors. The presentation draws attention to an underrepresented group and shows that older adults’ relationship to AI is not merely a technological issue, but also a social, economic, employment-related, and autonomy-related one.

Abstract: AI-Enabled Digital Twins in Consumer Ecosystems: Trust, Adoption, and the Future of Intelligent Digital Services

The rapid advancement of artificial intelligence (hereinafter, AI) is accelerating the integration of digital twins into consumer-facing digital ecosystems, fundamentally transforming the design and delivery of intelligent services. AI-enabled digital twins extend beyond static digital representations, evolving into dynamic, continuously learning entities capable of modeling, simulating, and optimizing interactions across platforms and services. Their growing adoption across industries, including retail, healthcare, smart environments, and digital platforms, signals a shift toward highly adaptive, data-driven ecosystems. This topic examines the role of AI-enabled digital twins in shaping consumer-oriented digital ecosystems, with a particular focus on trust formation, technology adoption, and system- level acceptance. The study conceptualizes digital twins as intermediaries between consumers and intelligent systems, facilitating personalized yet scalable interactions. In this context, trust emerges as a multi-dimensional construct influenced by perceived transparency, algorithmic explainability, data governance, and system reliability. The research develops a conceptual framework that integrates key dimensions such as digital trust, perceived risk, technological readiness, and ecosystem-level value co-creation. The framework highlights how AI-enabled digital twins contribute to the orchestration of intelligent services while simultaneously introducing new layers of complexity related to privacy, ethical design, and user autonomy. Particular attention is given to the tension between hyper- personalization and perceived surveillance, as well as the need for responsible AI practices. Methodologically, the study builds on interdisciplinary insights from artificial intelligence, marketing research, and digital innovation, combining conceptual analysis with emerging empirical evidence from digital service environments. The findings emphasize that the success of digital twin integration depends not only on technological sophistication but also on the ability of organizations to design transparent, trustworthy, and user-aligned systems. The presentation contributes to the growing body of research on AI and digital ecosystems by offering both theoretical advancements and practical implications. It provides strategic guidance for organizations seeking to leverage digital twins in consumer contexts, while also informing policymakers on the importance of governance frameworks that ensure ethical and sustainable deployment of AI-driven technologies.