Selected work

Representative Papers

HPC-Enabled Video-based Coastal Wave Parameter Estimation Using V-JEPA and Deep Spatiotemporal Learning

Jul 13, 2026

This study addresses the challenges of high cost, limited spatial coverage, and susceptibility to adverse weather that plague conventional in situ coastal wave observation methods, which hinder efficient acquisition of wave parameters. To overcome these limitations, the authors propose a deep learning framework that leverages monocular coastal video to jointly estimate five key wave parameters—significant wave height, maximum wave height, peak period, zero-crossing period, and wave direction—under data-scarce conditions. The approach integrates a self-supervised V-JEPA vision transformer, a SlowFast dual-stream temporal encoder, and Farneback optical flow, while incorporating Airy dispersion relation constraints to enforce physical consistency. Validated with only six annotated scenes and accelerated via high-performance computing, the model achieves Pearson correlation coefficients ranging from 0.451 to 0.832 across all five parameters, demonstrating both feasibility and strong cross-site generalization capability.

0 citationsRead paper

Physics-Guided Spatiotemporal Learning for Coastal Wave Peak Period Estimation from Video

Jun 11, 2026

This study addresses the limitations of conventional nearshore wave monitoring—high cost and limited spatial coverage—and the lack of physical interpretability in existing video-based deep learning approaches. The authors propose a physics-informed spatiotemporal deep learning framework that directly estimates the dominant wave period from video data by integrating automated temporal-variance-based region-of-interest detection, a hybrid Transformer–recurrent convolutional network, simulation-to-real transfer learning, and physics-informed regularization. The method balances predictive accuracy with fluid dynamic consistency: the Transformer architecture achieves optimal instantaneous precision, while the lightweight recurrent convolutional model demonstrates superior temporal stability and operational suitability. Physics-based regularization effectively suppresses non-physical solutions, and attention mechanisms selectively focus on hydrodynamically active surf-zone regions.

0 citationsRead paper

Building Capacity for Artificial Intelligence in Africa: A Cross-Country Survey of Challenges and Governance Pathways

Dec 05, 2025

AI education and talent development in Africa face structural barriers—including inequitable resource distribution, inadequate infrastructure, chronic underfunding, and policy gaps—that impede equitable and sustainable advancement. This study employs a cross-national survey of universities and industry stakeholders across five African countries, complemented by interdisciplinary empirical analysis. It identifies a critical paradox: high AI awareness coexists with low practical engagement. To address this, the paper proposes an “Inclusive AI Governance Framework,” centering on industry–university internship partnerships, targeted support for marginalized groups, and institutionalized work-integrated learning. Results indicate that curriculum relevance to labor-market needs significantly enhances competency readiness; however, economic constraints and infrastructural deficits remain primary bottlenecks. Strengthening industry–academia collaboration demonstrably narrows the skills gap, offering a scalable governance model for AI capacity building in the Global South.

0 citationsRead paper

Enriching Moral Perspectives on AI: Concepts of Trust amongst Africans

Aug 18, 2025

Current AI trust research is heavily grounded in WEIRD (Western, Educated, Industrialized, Rich, Democratic) sociocultural frameworks, resulting in the systematic marginalization of African perspectives. Method: This study conducts the first systematic investigation of how AI practitioners and researchers across Africa conceptualize trust, drawing on a mixed-methods survey—comprising quantitative questionnaires and qualitative in-depth interviews—with 157 professionals from 25 African countries, analyzed through sociological theory. Contribution/Results: We identify that AI trust in African contexts is fundamentally relational, shaped by communal ties, intergenerational knowledge transmission, and transnational practice. We propose “Afro-relationalism” as a novel theoretical framework, extending core trust constructs—including reliability and dependence—to foreground collective responsibility, context-sensitive accountability, and moral pluralism. Compared to WEIRD paradigms, African conceptions exhibit greater epistemic caution toward high-stakes AI applications. These findings challenge Western-centric trust models and provide empirically grounded, locally situated ethical foundations for global AI governance.

0 citationsRead paper

Recent publications

Latest Papers

HPC-Enabled Video-based Coastal Wave Parameter Estimation Using V-JEPA and Deep Spatiotemporal Learning

Jul 13, 2026

This study addresses the challenges of high cost, limited spatial coverage, and susceptibility to adverse weather that plague conventional in situ coastal wave observation methods, which hinder efficient acquisition of wave parameters. To overcome these limitations, the authors propose a deep learning framework that leverages monocular coastal video to jointly estimate five key wave parameters—significant wave height, maximum wave height, peak period, zero-crossing period, and wave direction—under data-scarce conditions. The approach integrates a self-supervised V-JEPA vision transformer, a SlowFast dual-stream temporal encoder, and Farneback optical flow, while incorporating Airy dispersion relation constraints to enforce physical consistency. Validated with only six annotated scenes and accelerated via high-performance computing, the model achieves Pearson correlation coefficients ranging from 0.451 to 0.832 across all five parameters, demonstrating both feasibility and strong cross-site generalization capability.

0 citationsRead paper

Physics-Guided Spatiotemporal Learning for Coastal Wave Peak Period Estimation from Video

Jun 11, 2026

This study addresses the limitations of conventional nearshore wave monitoring—high cost and limited spatial coverage—and the lack of physical interpretability in existing video-based deep learning approaches. The authors propose a physics-informed spatiotemporal deep learning framework that directly estimates the dominant wave period from video data by integrating automated temporal-variance-based region-of-interest detection, a hybrid Transformer–recurrent convolutional network, simulation-to-real transfer learning, and physics-informed regularization. The method balances predictive accuracy with fluid dynamic consistency: the Transformer architecture achieves optimal instantaneous precision, while the lightweight recurrent convolutional model demonstrates superior temporal stability and operational suitability. Physics-based regularization effectively suppresses non-physical solutions, and attention mechanisms selectively focus on hydrodynamically active surf-zone regions.

0 citationsRead paper

Building Capacity for Artificial Intelligence in Africa: A Cross-Country Survey of Challenges and Governance Pathways

Dec 05, 2025

AI education and talent development in Africa face structural barriers—including inequitable resource distribution, inadequate infrastructure, chronic underfunding, and policy gaps—that impede equitable and sustainable advancement. This study employs a cross-national survey of universities and industry stakeholders across five African countries, complemented by interdisciplinary empirical analysis. It identifies a critical paradox: high AI awareness coexists with low practical engagement. To address this, the paper proposes an “Inclusive AI Governance Framework,” centering on industry–university internship partnerships, targeted support for marginalized groups, and institutionalized work-integrated learning. Results indicate that curriculum relevance to labor-market needs significantly enhances competency readiness; however, economic constraints and infrastructural deficits remain primary bottlenecks. Strengthening industry–academia collaboration demonstrably narrows the skills gap, offering a scalable governance model for AI capacity building in the Global South.

0 citationsRead paper

Enriching Moral Perspectives on AI: Concepts of Trust amongst Africans

Aug 18, 2025

Current AI trust research is heavily grounded in WEIRD (Western, Educated, Industrialized, Rich, Democratic) sociocultural frameworks, resulting in the systematic marginalization of African perspectives. Method: This study conducts the first systematic investigation of how AI practitioners and researchers across Africa conceptualize trust, drawing on a mixed-methods survey—comprising quantitative questionnaires and qualitative in-depth interviews—with 157 professionals from 25 African countries, analyzed through sociological theory. Contribution/Results: We identify that AI trust in African contexts is fundamentally relational, shaped by communal ties, intergenerational knowledge transmission, and transnational practice. We propose “Afro-relationalism” as a novel theoretical framework, extending core trust constructs—including reliability and dependence—to foreground collective responsibility, context-sensitive accountability, and moral pluralism. Compared to WEIRD paradigms, African conceptions exhibit greater epistemic caution toward high-stakes AI applications. These findings challenge Western-centric trust models and provide empirically grounded, locally situated ethical foundations for global AI governance.

0 citationsRead paper