In an inspiring testament to scientific innovation and interdisciplinary collaboration, the Universidad Carlos III of Madrid (UC3M) has secured a remarkable achievement in the prestigious 12th edition of the BBVA Foundation’s Leonardo Grants for Researchers and Cultural Creators. Out of 1,623 proposals spanning ten distinct knowledge areas, three cutting-edge projects from UC3M emerged victoriously, each demonstrating groundbreaking potential in the realms of computer science, social sciences, and engineering. These grants, strategically targeted to mid-career researchers, underscore a keen commitment to fostering avant-garde approaches and transformative ideas poised to influence scientific and cultural landscapes on a global scale.
One of the most technically ambitious endeavors comes from David Martínez Rubio of UC3M’s Department of Signal Theory and Communications. His initiative, named ADOPT-AI (Adaptive optimization for AI), ventures into the core challenges of optimizing artificial intelligence model training. This research addresses the computational and temporal burdens often associated with tuning AI training algorithms. ADOPT-AI aims to develop an advanced meta-algorithmic framework capable of dynamically selecting and harmonizing multiple optimization algorithms. By effectively merging the advantages of each algorithm, this meta-optimization strategy promises to deliver unprecedented speed and efficiency, circumventing the need for prior knowledge of which algorithm fits best in a given context. The implications for large-scale AI models, including expansive language models, are profound: faster, more cost-effective AI training with the possibility for widespread deployment.
Parallel to this technological frontier, Amuitz Garmendia Madariaga from the Department of Social Sciences at UC3M leads the DRIFT project (Decentralization, Reforms, Institutions, and the Formation of Territorial Preferences). This research delves into the intricate mechanisms shaping citizens’ territorial preferences, scrutinizing how individual and collective attitudes toward the distribution of power within nation-states evolve. Employing a multidisciplinary methodology that blends political theory, historical contextualization, and empirical public opinion data, DRIFT investigates the foundational forces driving territorial and institutional transformations. By elucidating the subtle interplay between decentralization policies, reform processes, and citizen engagement, the project aspires to expand theoretical paradigms about democracy’s spatial dimensions and territorial governance.
Meanwhile, in the field of engineering, Manuel Torres Carrasco spearheads the BIO3D-TES project (Bio-inspired functionalization in sustainable 3D-printable systems for large-scale energy storage), housed within the Department of Materials Science and Engineering and Chemical Engineering. This visionary project is situated at the intersection of biomimetics, sustainable material science, and additive manufacturing. Inspired by efficient natural architectures such as spider webs and shark skin, BIO3D-TES seeks to fabricate smart thermal energy storage devices through innovative 3D printing techniques using environmentally sustainable and low-carbon materials. The bespoke design of these thermal batteries aims for superior heat transfer capabilities, adaptability to fluctuating thermal loads, and a systemic reduction in carbon footprint. Such bio-inspired designs have the potential to revolutionize energy storage infrastructures and substantially contribute to industrial decarbonization efforts worldwide.
These projects collectively attest to the breadth and depth of UC3M’s research prowess, reflecting a balanced embrace of theoretical complexity and practical application. The Leonardo Grants, each providing up to €50,000 in funding for a developmental period typically spanning 12 to 18 months, catalyze rapid advancements in these fields. The remarkably selective nature of the program—only 59 grants awarded from over 1,600 applicants—speaks volumes about the excellence and innovation embedded in these UC3M proposals. Since their inception in 2014, the Leonardo Scholarships have symbolized the BBVA Foundation’s enduring investment in fostering scientific and creative excellence.
Concerning ADOPT-AI, the technical underpinnings lie in adaptive meta-learning algorithms that are engineered to overcome the traditional pitfalls of gradient-based optimization methods prevalent in neural network training. By devising a hierarchy that intelligently switches or combines optimizer routines based on performance metrics and data characteristics during training, the project tackles overfitting risks and computational inefficiencies. Notably, this adaptive approach could facilitate the deployment of AI in more resource-constrained environments, broadening accessibility beyond dominant tech hubs.
The DRIFT project’s approach is marked by its innovative integration of heterogeneous data sources, historical institutional analyses, and the application of formal political models to infer preference evolution. By quantitatively mapping shifts in territorial attitudes across various demographic cohorts and geopolitical contexts, Madariaga’s research offers nuanced insights into the dynamic reflexivity between institutional design and citizen behavior. This could have profound policy implications for regions grappling with decentralization debates, autonomy movements, or federal restructuring.
Meanwhile, BIO3D-TES’s material scientific strategy hinges on synthesizing nanostructured composites that emulate biological surface functionalities, such as hydrophobicity and thermal conductivity control. The use of 3D printing allows for intricate structural customization at multiple scales, optimizing both mechanical integrity and thermal responsiveness. By marrying eco-friendly feedstocks with biomimetic geometries, the project aims to produce scalable prototypes that outperform conventional thermal energy storage solutions in efficiency and sustainability, a key enabler for the energy transition.
The broad diversity of these projects reflects the interdisciplinary imperative inherent to contemporary challenges—leveraging artificial intelligence, social science insights, and material innovation to confront complex, systemic problems. Moreover, these initiatives exhibit a strong potential for real-world impact, ranging from accelerating AI research cycles, informing democratic institutional reforms, to advancing sustainable industrial technologies central to global climate agendas.
Notably, the Leonardo Grants’ focus on mid-career researchers fills a critical funding gap often undersupported in traditional grant structures. This nurturing of emerging scientific leaders allows innovative projects—often too risky or novel for early-career or late-stage funding—to flourish. UC3M’s triad of projects exemplifies the cutting-edge scholarship and creativity this model intends to cultivate.
As global society confronts rapidly evolving technological paradigms and socio-political transformations, research endeavors such as those at UC3M offer vital foresight. The adaptive optimization techniques envisioned in ADOPT-AI could redefine algorithmic efficiency across AI-dependent industries, while DRIFT’s exploration of territorial preference formation resonates in an era marked by political fragmentation and demands for decentralization. Simultaneously, BIO3D-TES’s bio-inspired thermal solutions align with urgent calls for sustainable energy systems capable of integrating seamlessly with renewable resources.
Ultimately, the combination of advanced theoretical modeling, innovative methodological frameworks, and applied technological design embodied by these projects delineates a robust roadmap for research that not only interrogates foundational scientific questions but also drives tangible, societally relevant progress. UC3M’s success in the BBVA Foundation’s Leonardo Grants represents a beacon of scholarly excellence and a compelling vision for the future of integrative research scholarship.
Subject of Research: Adaptive optimization in artificial intelligence training, territorial preference formation in democratic states, and biomimetic sustainable thermal energy storage.
Image Credits: UC3M
