Abstract

As intelligence becomes abundant, adaptive and increasingly capable of acting across organisational boundaries, the central challenge for institutions is no longer simply how to adopt artificial intelligence. It is how to govern it.

This keynote introduces ATMOS Global Fourth System™: a higher-order governance framework designed to sit above human intuition, human deliberation and increasingly autonomous forms of artificial intelligence. Systems 1 and 2 remain constrained by established human cognitive limits. System 3 extends decision capacity through scalable and increasingly self-evolving machine intelligence, yet introduces its own risks: opacity, dependency, model error, misalignment and accelerated systemic feedback.

The role of System 4 is therefore not to generate more intelligence. Its role is to govern the architecture within which intelligence is directed, constrained, evaluated and acted upon.

The framework is organised around three enduring responsibilities. Value defines what the system is ultimately for: its purpose, fiduciary and ethical obligations, human interests and non-negotiable principles. Coordination aligns people, AI systems, data, institutions and decision rights across increasingly complex networks. Continuity preserves institutional memory, resilience, accountability and long-term stewardship through technological change, leadership transition and disruption.

In this model, governance determines where authority resides, what may be delegated, what must remain under human responsibility, and how decisions remain legitimate as systems learn and evolve. The strategic implication is significant: as intelligence becomes more abundant, the scarce capability may increasingly be the capacity to decide what a system should optimise, under which constraints, and where accountability ultimately resides.

The keynote examines these questions through two high-stakes domains. In finance, ATMOS Global Fourth System™ applies to AI-supported capital allocation, advice, risk, compliance, market behaviour and intergenerational stewardship. In health, it applies to AI-enabled diagnosis, triage, care coordination, research and patient engagement, while preserving clinical accountability, patient dignity and public trust.

Across both sectors, the proposition is the same: enduring advantage will not lie merely in possessing superior intelligence, but in governing its purpose, coordinating its effects and preserving continuity of value across time.

ATMOS Global Fourth System™ calls for a new executive capability: leaders and boards able to govern self-evolving intelligence not as a technology project, but as an institutional, ethical and strategic responsibility.

Biography

Dr Orestis D. Valianatos is the Global President and CEO of ATMOS Global Pty Ltd (ATMOS Global ™), a highly accredited and accomplished professional company director, an astute investor in the capital markets, an established innovator, a role model, a visionary thinker with a diverse portfolio of formal qualifications (BSc, MSc, PhD in Atmospheric Physics, Climate & Sustainability, Master of Business Administration, Professional Doctorate in International Business and Sustainability, Corporate Governance training as a Professional Board Director, and soon a Diploma of Finance), experience, unique insight and influential global perspective across multiple domains including artificial intellignce, leadership, strategy and innovation, sustainability, ESG, climate-tech, thematic and sustainable investing. Entrepreneur and agent of change, he has authored more than 50 ground-breaking international research papers as a subject matter expert working in partnership with senior personnel from government departments, universities and major private clients from the mining, manufacturing, energy & utilities, health and finance sectors.

Abstract

This paper presents a comprehensive study on the use of Physics-Informed Neural Networks

(PINNs) for solving stiff initial value problems (IVPs) and stiff inverse problems. Two formulations are developed in full detail: the soft-constraint PINN and the hard-constraint

PINN. We then apply five stiff-oriented techniques—dynamic re-weighting

of residual terms, temporal domain decomposition (cPINN/XPINN), discrete-time PINN

with implicit Radau IIA, continuous adjoint methods for gradient computation, and discrete-time PINN for inverse problems—and compare the results. All variants are compared

against the classical fourth-order Runge–Kutta method (RK4) and the Dormand–Prince

adaptive step-size integrator on a stiff scalar IVP. Complete forward and backward

passes (summarized), pointwise and aggregated error metrics (E∞, L2, relative L2, MSE)

are provided. Errors are computed strictly at grid points, with no interpolation. For the stiff inverse problem, an unknown parameter θ in the ODE right-hand side must

be estimated from noisy synthetic observations. Strategies are compared: classical

soft PINN, classical hard PINN, stiff-oriented PINN, discrete-time PINN with implicit

Radau IIA, RK4 with uniform grid search, RK4 with gradient search using complex-step

finite differences (CSFD), Dormand–Prince with CSFD, and continuous adjoint gradient

descent. The Python implementation reproduces every table. The numerical results reveal qualitatively different behaviour from what is often claimed in the literature: (i) the classical soft PINN fails on the forward stiff IVP (E∞≈ 0.9998), while the hard PINN resolves the transient only to E∞ ≈ 1.493 (its performance is highly sensitive to the length of the Adam stage and to the local minimum reached); among continuous PINN formulations, the hard PINN combined with dynamic re-weighting achieves the best forward accuracy (E∞≈ 0.123); (ii) the discrete-time PINN with implicit Radau IIA achieves E∞≈ 0.8851 with a residual loss of 1.226×10−2, but still does not reach classical accuracy; (iii) in the stiff inverse problem, every method fails: the classical soft PINN lands at θ̂ ≈ 451.8345, the hard PINN collapses to θ̂ ≈ 37.1151, the stiff-oriented PINNs collapse further to θ̂ ≈ 28.6170 (soft) and 11.3904 (hard), the discrete-time PINN lands at θ̂ ≈ 489.4922, the RK4 grid search returns the upper boundary θ̂ = 2000.0000, and the CSFD-based Newton searches and the continuous adjoint all stall at the initial guess θ̂ = 500.0000. A sensitivity diagnostic makes explicit that the chosen observation design places the smallest non-zero observation time at t = 1/9, where the sensitivity |∂u/∂θ| = t e−θt ≈ 6.18 × 10−50 is far below the noise level σ = 0.01; the inverse problem is therefore practically unidentifiable independently of the optimizer. The complex-step gradient at θ = 500 is J ′= −4.4902 × 10−15 — essentially identical to the value obtained with hcs = 10−20, confirming that the true gradient is at or below the numerical floor. The closed-form continuous adjoint gradient at θ = 500 is −1.5236×10−7 (a sign that depends on

the Lagrangian convention), whose magnitude is five orders of magnitude smaller than the

noise-driven residuals. These findings are discussed honestly, and the practical recommendations are revised accordingly: for this class of extremely stiff problems (θ = 1000

with T = 1, i.e. θT = 1000 ≫ 1), neither continuous-time PINNs nor discrete-time PINNs

nor adjoint-based parameter estimation are adequate; adaptive classical integrators remain

the only reliable tool.

Biography

Associate Professor Dr Yeak Su Hoe is an academic at the Department of Mathematical Sciences, Faculty of Science, Universiti Teknologi Malaysia (UTM), Malaysia. He obtained his Ph.D. from Nanyang Technological University, Singapore, specialising in molecular dynamics and computational quantum mechanics. Dr Yeak's research interests encompass numerical analysis, numerical methods, scientific computation, mathematical modelling and simulation, molecular dynamics, computational quantum mechanics, and stochastic models. His research focuses on the development and application of computational and numerical techniques to address mathematical and scientific problems, with applications in the physical sciences and engineering. He leads the Industrial and Scientific Computation (ISC) Research Group at UTM, which undertakes research in numerical algorithms, scientific computing, mathematical modelling and simulation, and computational methods for interdisciplinary scientific and engineering applications. In addition to his research activities, he contributes to postgraduate education and academic programme coordination. Through his research, teaching and postgraduate supervision, Dr Yeak is committed to advancing applied and computational mathematics and promoting the application of mathematical methods to scientific and technological challenges.

Abstract

Abstract

Modern intrusion detection systems increasingly rely on data-driven methods to analyze large and heterogeneous network traffic. Association rule mining provides an interpretable mechanism for capturing recurring behavioral patterns, but the large number of generated rules often limits practical usability. In contrast, ensemble learning methods scale effectively but typically provide limited transparency. This paper presents a framework that combines FP-Growth Association Rule Mining with a hybrid Random Forest–ExtraTrees classifier to evaluate rule quality while preserving interpretability. Association rules are mined from network traffic corresponding to seven MITRE ATT&CK tactics combined with benign activity and labeled according to rule confidence. The resulting rule sets are used to train a hybrid ensemble model that distinguishes high-confidence from low-confidence rules using statistical and structural rule attributes. Experimental evaluation across datasets containing tens of thousands to millions of rules shows perfect separation between rule classes, with accuracy, precision, recall, F1-score, and AUC. Feature importance analysis indicates that FP-Growth confidence alone defines the decision boundary across all attack types. To support transparency, ensemble decision paths are mapped directly back to the original association rules, exposing the antecedent–consequent relationships that drive classification outcomes. This approach enables scalable rule validation while retaining semantic clarity. The results demonstrate that association-rule confidence provides a stable and attack-agnostic indicator of rule reliability and that hybrid ensemble models can be used as effective, explainable validation mechanisms in large-scale intrusion detection workflows. d very high accuracy, precision, recall, F1-score, and AUC measures, allowing for the explainability of rules with high confidence. This framework also allows for the successful classification of rare attacks, since FP-Growth association rule mining generates a large number of rules.

Biography

Dr. Sikha Bagui is Distinguished University Professor and Askew Fellow in the Department of Computer Science, at The University West Florida, Pensacola, Florida. Dr. Bagui is active in publishing peer reviewed journal articles mainly in the areas of Big Data Analytics, Machine Learning and AI. Dr. Bagui has worked on funded as well unfunded research projects and has 175+ peer reviewed publications in highly selected journals and conferences. She has also co-authored several books on Databases and SQL. Bagui also serves as Associate Editor and is on the editorial board of several journals.

Abstract

Collection and analysis of data from deployed networks is essential for understanding communication networks. Hence, data mining and statistical analysis of network data have been employed to determine traffic loads, analyze patterns of users' behavior,  predict future network traffic, and detect traffic anomalies. The Internet has historically been prone to failures and attacks that significantly degrade its performance, affect the Internet connectivity, and cause routing disconnections. Frequent cases of various cyber threats have been encountered over the years and, hence, detection of anomalous behavior is a topic of great interest in cybersecurity. In described case studies, traffic traces collected by various collection sites are used to classify network anomalies. Various anomaly and intrusion detection approaches based on machine learning have been employed to analyze collected data. Deep learning, broad learning, gradient boosted decision trees, and reservoir computing algorithms were used to develop models based on collected datasets that contain Internet worms, viruses, power outages, ransomware events, router misconfigurations, Internet Protocol hijacks, and infrastructure failures in times of conflict. The reported results indicate that while performance of machine learning models greatly depends on the used datasets, they are viable tools for detecting the Internet anomalies.

Biography

Ljiljana Trajkovic received the Dipl. Ing. degree from University of Pristina, Yugoslavia, the M.Sc. degrees in electrical engineering and computer engineering from Syracuse University, Syracuse, NY, and the Ph.D. degree in electrical engineering from University of California at Los Angeles. She is currently a professor in the School of Engineering Science, Simon Fraser University, Burnaby, British Columbia, Canada. Her research interests include communication networks and dynamical systems. Dr. Trajkovic served as IEEE Division X Delegate/Director, President of the IEEE Systems, Man, and Cybernetics Society, and President of the IEEE Circuits and Systems Society. She serves as Editor-in-Chief of the IEEE Transactions on Human-Machine Systems. She is a Distinguished Lecturer of the IEEE Systems, Man, and Cybernetics Society and was a Distinguished Lecturer of the IEEE Circuits and System Society. She is a Fellow of the IEEE.

Abstract

Good mixing matters everywhere; from blending fluids on a microchip to stirring patterns in the atmosphere, and in countless industrial processes. But how do you stir a fluid as efficiently as possible using only a handful of fixed devices? This talk looks at how to design the timing and motion of simple stirrers to achieve the best possible mixing.

We present two complementary approaches to this problem.

The first uses a carefully designed computational method that respects the physical laws governing the fluid; conserving mass and energy exactly, down to the precision of the computer. This matters because it lets us trust that any improvement in mixing we observe is real, and not just an artifact of numerical error. Using this approach, we show that simply optimizing when two simple, steady stirrers move  (without changing their design) can turn slow, inefficient mixing into a much faster process.

The second approach offers a shortcut. Rather than tracking the full complexity of the mixing fluid, we focus on a single measurable feature: how much a stretched interface between two fluids elongates over time. It turns out that stretching this boundary as much as possible is a reliable stand-in for good mixing overall, since a longer, more stretched interface means finer, more thoroughly blended filaments. This simplification allows for a much lighter computational approach (one that doesn't require a fine grid) while still achieving mixing performance comparable to (or better than) the more computationally expensive first method, at a fraction of the cost (up to twenty times faster).

Together, these results show that achieving trustworthy, efficient mixing requires both computational methods that respect the underlying physics and simplified geometric descriptions that capture the essence of how fluids stretch and fold.

Biography

Enrique Zuazua (Eibar, Basque Country–Spain, https://dcn.nat.fau.eu/enrique-zuazua/) holds, since September 2019, the Chair for Dynamics, Control, Machine Learning and Numerics - Alexander von Humboldt Professorship, at the Department of Mathematics of the Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) in Germany and part-time appointments at Universidad Autónoma de Madrid (UAM) and the University of Deusto, Bilbao, Spain. He is also a member of the Basque Academy “Jakiunde“, Fellow of the Artificial Intelligence Industry Academy (AIIA), of the Asia-Pacific Artificial Intelligence Association (AAIA), and of the Academia of Europaea, and scientific advisor of the artificial intelligence company Sherpa AI in Bilbao. He holds a degree in Mathematics (1984) from the University of the Basque Country, and a dual Ph.D. degree from the same university (1987) and the Université Pierre et Marie Curie, Paris (1988). In 1990 he became Professor of Applied Mathematics at the Universidad Complutense de Madrid, to later move to UAM in 2001. His fields of expertise in the broad area of Applied Mathematics include Partial Differential Equations, Systems Control and Numerical Analysis and Machine Learning. He has been awarded the Euskadi (Basque Country) Prize for Science and Technology 2006 and the National Julio Rey Pastor Prize 2007 in Mathematics and Information and Communication Technology and the Advanced Grants of the European Research Council (ERC) NUMERIWAVES in 2010, DYCON in 2016 and CoDeFeL in 2022. In 2022 he was awarded the W.T. and Idalia Reid Prize of SIAM. He is an invited section speaker in “Control and Optimization” in ICM2006, Madrid. With over 300 articles published, his work had an important impact (h-index = 50). He has supervised 30 PhD students and a broad network of master students, post-doctoral researchers and research and management technicians. He is the co-editor-in-chief of the Journals “Mathematical Control and Related Fields” and “Advances in Continuous and Discrete Models” and member of the editorial committee of other Journals and of scientific committees of various centers and agencies. He was the first Manager for Mathematics of the Spanish National Research Plan in 1999-2002, the Founding Scientific Director of the Basque Center for Applied Mathematics (BCAM) in 2008-2012 and of the Chair of Computational Mathematics at the University of Deusto-Bilbao in 2016, both in Bilbao. Since 2021 he is the inaugural Speaker of the FAU Research Center for Mathematics of Data (MoD). He also develops an intense dissemination agenda, gathered at https://cmc.deusto.eus/enzuazua/

Abstract

Blackholes, Galaxies, particles such as photons, neutrons, Protons, also waves having

important properties that is interresting and valuable for us to measure in a comprehensive , analogue and coherent way , often necceary to have a good pinpointed value, but craving this rigid and commanding and mabe also neccesicite results do take a lot of conditional demands with their call,.. We must follow principles of logic analogy and derivational theorems, also having emphasis on orthogonality,.. for beeing able the imposition of any proof.

The problem is solved by connecting the properties in a strong derivational chain.

This then can lead us to a complete theory.

Biography

Tony Barrera is a certified autodidact math genius. He have published more than 42 Ordinary high rated scientific papers And up to several hundred publications ,computer simulations and animations In different subjects, scientific papers in mathematics , computer graphics, numerical analysis, astrophysics and Particle Atomic physics. Tony does research general together with prof Ewert Bengtsson, Prof Anders Hast and Physicist Bo Thelin and the crew of Barrera Science Lab. Tony Barrera have been working for the company AB Consonant with implementation of the Fast Fourier Transform, FFT , and as a computergrahics researcher at Cycore AB (Webbgraphics). Constructor of about 10 - 20 different Graphics Engines , Assembler ,Basic, Pascal and/or C++ computer languages. Started with mahematics and science at age 3, got the first Astronomy book (from mother Elisabeth Mercadal) at 9, the first Chemistry set at 10 , Chief (head) of Barrera Science Lab.

Abstract

This talk will present experimental data pointing to the discovery of room-temperature multiferroicity in doped polyaniline synthesized under an applied magnetic field, which is enhanced via Cu2+-doping. It is claimed that while the applied magnetic field stretches and aligns the polymeric chains Cu2+-doping binds together lateral chains, likely via a nitrogen-coordinated complex. Experimental evidence of the effect of the applied magnetic field on the crystallinity of the synthesized polymer will be provided. Likewise, experimental evidence of the Cu2+-complexing will be presented. Room-temperature switchable magnetization and polarization hysteresis cycles, reaching impressive magnetization and polarization at saturation, with strong electric exchange bias, will be presented and discussed. Real and imaginary parts of permeability and permittivity, exhibiting resonance features at about the same position, strongly suggests the onset of magnetoelectricity. A proposed model picture, accounting for both ferroelectricity and exchange bias, assumes ferromagnetic and ferroelectric orders within the same nanodomain, which are represented by crystalline polyaniline nano-islands immersed within the amorphous polyaniline template.

Biography

Professor De Morais (H63), was full Professor of Physics at the University of Brasilia (UnB) – Brazil up to 2013. UnB’s Emeritus Professor (2014); Visiting Professor at the Huazhong University of Science and Technology – China (2012-2015); Distinguished Professor at the Anhui University – China (2016-2019); Professor at the Catholic University of Brasília – Brazil (since 2018); CNPq-A Research Fellow since 2010; 2007 Master Research Prize from UnB. Two-years (1987-1988) post-doc position with Bell Communications Research, New Jersey – USA. PhD in Solid State Physics (1986) from the Federal University of Minas Gerais – Brazil. More than 14,500 citations. More than 500 papers (Web of Science). About 300 international invited talks (35 countries), and 18 filed patents.

Abstract

Nanomedicine is the use of nanomaterials to improve disease prevention, detection, and treatment which has resulted in hundreds of FDA approved medical products. While nanomedicine has been around for several decades, new technological advances are pushing its boundaries. For example, this presentation will present an over 30 year journey of commercializing nanomaterials for orthopedic implants now in over 45,000 patients to date showing no signs of failure. Current orthopedic implants face a failure rate of 5 – 10% and sometimes as high as 60% for bone cancer patients. Further, this talk will present future research directions into using atomic layer deposition (ALD) to create such nanostructures on implants to reduce infection and improve bone growth. Sensors grown off of orthopedic implants using ALD will also be discussed in which cell presence on orthopedic implants can be detected and quantified. Such information can also be communicated to a handheld device to better inform surgeons on chances of implant success or failure. Such sensors can also release pharmaceutical agents and/or nanoparticles on-demand to ensure implant success. Lastly, this talk will cover how Artificial Intelligence (AI) can be combined into today’s orthopedic implants to predict implant success or failure in the years that follow.

Biography

Thomas J. Webster’s (H index: 139) degrees are in chemical engineering from the University of Pittsburgh (B.S., 1995; USA) and in biomedical engineering from RPI (Ph.D., 2000; USA). He has formed over a dozen companies who have numerous FDA approved medical products currently improving human health in over 45,000 patients. He is currently helping those companies and serves as a professor at Brown University, Saveetha University, Hebei University of Technology, UFPI, University of the Basque Country, and others. Dr. Webster has numerous awards including: 2020, World Top 2% Scientist by Citations (PLOS); 2020, SCOPUS Highly Cited Research (Top 1% Materials Science and Mixed Fields); 2021, Clarivate Top 0.1% Most Influential Researchers (Pharmacology and Toxicology); 2022, Best Materials Science Scientist by Citations (Research.com); and is a fellow of over 8 societies. Prof. Webster is a former President of the U.S. Society for Biomaterials and has over 1,350 publications to his credit with over 55,000 citations. He was recently nominated for the Nobel Prize in Chemistry. Prof. Webster also recently formed a fund to support Nigerian student research opportunities in the U.S.

Abstract

Aging is accompanied by a progressive decline in skeletal regenerative capacity, driven largely by shifts in bone marrow-derived mesenchymal stem cell lineage commitment that favor adipogenesis over osteogenesis. Overcoming this clinical bottleneck requires robust bioanalytical approaches capable of translating molecular signatures into functional interventions.
Our research address highlights how high-resolution, multiplexed mass spectrometry serves as a powerful applied biosciences platform to decode the complex signaling, metabolic, and gene regulatory dysfunctions inherent in cellular aging. Drawing from integrative tandem mass tag-based phosphoproteomic and total proteomic profiling of murine stem cell models, this presentation demonstrates how systems-level analytics isolate over 500 significantly altered proteins and phosphorylation events. Attendees will explore how data-driven workflows identify critical bioenergetic and regulatory shifts, including deregulated clathrin-mediated endocytosis, nucleocytoplasmic transport, pathway-specific signaling aberrations like ERK-1/2 hyperactivation, and transcriptional dysregulation. By leveraging quantitative proteomics, this talk bridges analytical mass spectrometry and regenerative medicine, establishing a translational foundation for designing targeted peptide therapeutics and bioengineering strategies to combat age-associated osteoporosis.

Biography

Dr. Anja Nohe is an Associate Professor in the Department of Biological Sciences at the University of Delaware, whose scientific career bridges molecular cell biology, quantitative biophysics, and systems-level cellular dynamics. Earning her doctorate in Chemistry from the University of Würzburg in Germany and completing postdoctoral research at the University of Western Ontario, Dr. Nohe has established a robust academic trajectory examining the fine-grained physical rules governing signal transduction. Her research program heavily intersects with systems biology by investigating how microscopic protein interactions and nanoscale receptor dynamics scale up to govern complex cellular fates, such as stem cell lineage commitment, osteogenesis, and musculoskeletal health. Rather than viewing cellular pathways as static linear diagrams, her laboratory applies quantitative biophysical approaches—such as the Family of Image Correlation Spectroscopy (FICS) and advanced fluorescence microscopy—to map out spatial-temporal protein dynamics, clustering, and membrane trafficking in real time in live cells. Moreover, her laboratory heavily integrates systems biology and bioinformatics approaches to understand how cellular dynamics change over time, particularly focusing on aging, bone marrow-derived mesenchymal stem cells (BMSCs), and age-related conditions like osteoporosis and osteoarthritis. As organisms age, stem cells frequently experience desensitization and altered lineage commitment—shifting away from osteoblastogenesis toward adipogenesis and bone marrow fat accumulation. Dr. Nohe's group applies quantitative modeling, phosphoproteomics, and transcriptomic analyses to map out how aging systematically reprograms signaling cascades, metabolic networks, and gene regulatory dynamics.

Abstract

In the face of escalating cyber threats, the fusion of artificial intelligence (AI) with cybersecurity strategies has become a cornerstone for modern defense systems. This presentation, delivered by Dr. Farah Jemili, explores the pivotal role of generative AI in fortifying cybersecurity frameworks. As modern communication technologies generate vast amounts of data daily, the presentation addresses the pressing need to harness this data to safeguard Industry 4.0 infrastructures, which are increasingly vulnerable to cyber-attacks. With cyber-attack damages projected to reach $8 trillion in 2024 and escalate to $10.5 trillion by 2025, innovative solutions are essential.The presentation outlines the current cybersecurity landscape, identifying key challenges such as data collection, storage, processing, and the accurate detection of cyber threats. It emphasizes the transformative potential of AI methodologies, including machine learning, deep learning, and generative learning, in developing robust intrusion detection systems and real-time threat response mechanisms.A detailed examination of real-world AI applications in cybersecurity is provided, showcasing technologies like Cylance for malware prevention, AEG for automatic exploit generation, AI2 for predictive threat analysis, and IBM’s Watson for IoT network analytics. The comparative study of various deep learning models highlights their distinct advantages and limitations in cybersecurity contexts.Furthermore, the presentation discusses the integration of AI with other technological pillars such as   cloud  computing   and   big  data,   illustrating   how  these   synergies   enhance  cybersecurity capabilities. The research contributes  to  understanding how AI  can be leveraged  to develop intelligent, adaptive, and resilient cybersecurity systems, ultimately aiming to mitigate the risks and impacts of cyber-attacks.The presentation  concludes with  perspectives on future research  directions and  the evolving landscape of AI in cybersecurity, underscoring the critical role of continuous innovation and collaboration in this domain. 

Biography

Farah JEMILI is an Associate Professor at the Higher Institute of Computer Science and Communication Technologies of Sousse (ISITCOM), Tunisia. She has completed her Ph.D. in 2010, from the National School of Computer Sciences (ENSI), Tunisia. She is a Scientific Referent at the Tunisian Agency of Evaluation and Accreditation (ATEA), and Internship Director at ISITCOM. She has been member of the Scientific Council of ISITCOM for 3 years (2011-2014), and Head of the Department of Computer Science at ISITCOM for 3 years (2017-2020). Her research interests include Artificial Intelligence, Cyber Security and Big Data Analysis. She served as a Reviewer for many international conferences and journals. She has published around 60 Research papers in international journals and conferences and has presented many invited and contributed Talks at international conferences.

Abstract

In recent years, hybrid functions by combining block-pulse functions with orthogonal polynomials or polynomial series have found their way into many fields of science and engineering. Many researchers have used hybrid functions to analyze problems with greater computational complexity, and they have proven to be powerful tools for exploring new directions in solving scientific and engineering problems.

A wavelet is a numerical tool for solving many kinds of problems. Wavelets have many attractive features, including arbitrary regularity and good localization. After wavelets were successfully applied to signal and image processing in 1980, wavelet analysis became important. They have created a common link among mathematicians, physicists, and electrical engineers. Indeed, "wavelets" are a versatile tool with rich mathematical content and great potential for applications.

Fractional differential equations (FDEs) are generalizations of ordinary differential equations to an arbitrary (non-integer) order. FDEs have attracted increasing attention because they can model complex phenomena. Fractional derivatives are used in physics, mechanics, control, aerospace, and so on. Generally, most FDEs do not have exact analytic solutions. Therefore, finding numerical solutions is important. In this talk, we apply hybrid functions and wavelets to develop an efficient numerical method for FDEs. We compare the numerical solutions with available exact or approximate solutions to assess the accuracy of the proposed method.

Biography

Professor Mohsen Razzaghi received his Bachelor's in Mathematics from the University of Sussex in England, his Master's in Applied Mathematics from the University of Waterloo in Canada, and his doctorate from the University of Sussex. He is currently a professor in the Department of Mathematics and Statistics at Mississippi State University (MSU) in the USA and served as Head of the Department from 2006 to 2025. Dr. Razzaghi was awarded two prestigious Fulbright fellowships for the academic years 2011-2012 and 2015-2016. During each period, he spent 9 months at the Technical University of Civil Engineering in Bucharest, Romania, one of Europe’s most prestigious universities. In addition, he received two Fulbright Specialist Awards: one at the University of Bucharest in Romania in 2019 and another at Prince of Songkhla University in Thailand in 2024. Dr. Razzaghi was named a W.L. Giles Distinguished Professor at MSU in 2019. The highest award the University can bestow upon a faculty member. He also received the Robert Wolverton Legacy Award in 2024, the highest honor the College of Arts & Sciences bestows on those who have made sustained contributions to the goals of the liberal arts & sciences. His research area is Applied and Computational Mathematics, encompassing the development, analysis, and implementation of mathematical models and numerical methods in the sciences and engineering. He is particularly interested in optimal control, fractional calculus, orthogonal functions, and wavelets. According to Scopus, he has over 280 refereed publications in high-quality international journals across applied mathematics, mathematical physics, and engineering. In addition, according to Scopus, he has over 7,300 citations across 4,445 documents, with an h-index of 44, and one of his papers, coauthored with a Ph.D. student at MSU, has been cited over 735 times.

Abstract

The ability to construct scientific arguments requires students to distinguish observable evidence from scientific concepts and use both to support claims. However, students often struggle to differentiate evidence from reasoning, particularly when reasoning requires connecting observations to broader scientific principles. This presentation introduces a structured approach to Claim, Evidence, and Reasoning (CER) assessment, featuring rubrics and templates designed to support student development from sixth through eighth grade.

The approach uses a consistent assessment structure across grade levels while progressively increasing expectations for scientific thinking. In sixth grade, students focus on distinguishing evidence from reasoning and connecting observations to key scientific concepts. In seventh grade, expectations expand to include quantitative measurements, multiple sources of evidence, and evaluation of how data were collected. By eighth grade, students are expected to evaluate competing claims, use evidence to support and refute explanations, and incorporate both disciplinary core ideas and crosscutting concepts into their reasoning.

The presentation demonstrates how these expectations can be translated into explicit rubric criteria aligned with the Next Generation Science Standards (NGSS) Science and Engineering Practices. Evidence cards, color-coded templates, and scaffolded writing supports help students organize scientific information while gradually developing independence. A central principle of this approach is that scientific reasoning involves more than restating evidence: students must explain how scientific concepts connect observations to a claim.

This framework offers science educators a practical method for designing transparent, developmentally appropriate assessments that strengthen scientific argumentation, support meaningful feedback, and promote progression toward more sophisticated scientific thinking.

Keywords: Claim, Evidence, and Reasoning (CER); science assessment; scientific argumentation; NGSS; middle school science; rubric design; scientific reasoning.

Biography

**Andrew McLaren** is a science and mathematics educator, curriculum designer, and educational consultant with experience teaching in public schools in Oregon and California. Through his work with *McLearn with McLaren*, he develops NGSS-aligned science assessments, rubrics, and professional learning resources to support effective science instruction. He has designed assessments for all eight NGSS Science and Engineering Practices and provides professional development to help educators strengthen student learning through inquiry, scientific reasoning, and evidence-based practices. He has developed rubrics for CER, which are used in multiple districts now. Andrew also contributes to science education through curriculum review, consulting, and educational content creation. He currently works as a Professional Learning Specialist with Amplify Desmos Math, supporting educators in improving teaching and learning.