Institution profile

Centre for Research and Technology Hellas

Academic institutioneurope · gr
Official website
Research library39linked papers
Opportunities0open roles
Selected work

Representative Papers

Three Necessary Principles for Self-Supervised Visual Representation Learning

Aug 08, 2026

This work addresses the lack of a unified theoretical foundation in unsupervised visual representation learning, where existing methods struggle to simultaneously achieve semantic invariance, spatial structure modeling, and non-degenerate solutions. The authors propose three essential principles—observation, prediction, and regularization—and formulate them within a unified energy-based decomposition framework, offering the first formalization of core self-supervised learning criteria. Through rigorous analysis of gradient complementarity, convergence guarantees for momentum encoders, and a negative-sample-free alignment theory, the study exposes fundamental limitations of contrastive learning and momentum mechanisms, demonstrating that all three principles are indispensable. Controlled experiments, including block retrieval evaluations, confirm that optimal performance is attained only when these principles operate in concert.

0 citationsRead paper

AI-Farol: Co-Evolutionary Dynamics in a Multi-Agent Two-Sided Learning Framework

Aug 05, 2026

This work addresses the limitations of the classical El Farol bar problem, which assumes a passive venue and full observability—conditions ill-suited for real-world uncertainty and resource coordination challenges. The authors propose a bilateral learning framework in which the bar is modeled as an active AI-driven mechanism designer operating under partial observability. Agents learn attendance strategies based on incomplete information, while the bar dynamically adjusts pricing to jointly optimize revenue, utilization, and sustainability. This approach uniquely treats the venue as a strategic, learning-capable participant, integrating multi-agent reinforcement learning with adaptive mechanism design. The resulting co-evolution of agents and institutional rules establishes a novel paradigm for resource coordination in complex adaptive systems.

0 citationsRead paper

Scaling Vision-Language Models Is Not Enough to Mitigate Bias

Jul 30, 2026

This study addresses the limited robustness of large vision-language models against spurious correlations—such as single- and multi-attribute biases—and the unclear relationship between model scale and debiasing efficacy. Through a systematic empirical analysis of 194 publicly available models, the authors evaluate how model size, training data, and architectural design influence bias sensitivity on ImageNet, CelebA, and UrbanCars benchmarks. They find that increasing model scale has nearly no effect on mitigating complex multi-attribute biases (ρ = 0.05), whereas high-quality, large-scale training data consistently improves worst-group accuracy by up to 25%. The impact of architectural choices, however, is highly dependent on the specific bias type and its spatial distribution.

0 citationsRead paper
Recent publications

Latest Papers

Three Necessary Principles for Self-Supervised Visual Representation Learning

Aug 08, 2026

This work addresses the lack of a unified theoretical foundation in unsupervised visual representation learning, where existing methods struggle to simultaneously achieve semantic invariance, spatial structure modeling, and non-degenerate solutions. The authors propose three essential principles—observation, prediction, and regularization—and formulate them within a unified energy-based decomposition framework, offering the first formalization of core self-supervised learning criteria. Through rigorous analysis of gradient complementarity, convergence guarantees for momentum encoders, and a negative-sample-free alignment theory, the study exposes fundamental limitations of contrastive learning and momentum mechanisms, demonstrating that all three principles are indispensable. Controlled experiments, including block retrieval evaluations, confirm that optimal performance is attained only when these principles operate in concert.

0 citationsRead paper

AI-Farol: Co-Evolutionary Dynamics in a Multi-Agent Two-Sided Learning Framework

Aug 05, 2026

This work addresses the limitations of the classical El Farol bar problem, which assumes a passive venue and full observability—conditions ill-suited for real-world uncertainty and resource coordination challenges. The authors propose a bilateral learning framework in which the bar is modeled as an active AI-driven mechanism designer operating under partial observability. Agents learn attendance strategies based on incomplete information, while the bar dynamically adjusts pricing to jointly optimize revenue, utilization, and sustainability. This approach uniquely treats the venue as a strategic, learning-capable participant, integrating multi-agent reinforcement learning with adaptive mechanism design. The resulting co-evolution of agents and institutional rules establishes a novel paradigm for resource coordination in complex adaptive systems.

0 citationsRead paper

Scaling Vision-Language Models Is Not Enough to Mitigate Bias

Jul 30, 2026

This study addresses the limited robustness of large vision-language models against spurious correlations—such as single- and multi-attribute biases—and the unclear relationship between model scale and debiasing efficacy. Through a systematic empirical analysis of 194 publicly available models, the authors evaluate how model size, training data, and architectural design influence bias sensitivity on ImageNet, CelebA, and UrbanCars benchmarks. They find that increasing model scale has nearly no effect on mitigating complex multi-attribute biases (ρ = 0.05), whereas high-quality, large-scale training data consistently improves worst-group accuracy by up to 25%. The impact of architectural choices, however, is highly dependent on the specific bias type and its spatial distribution.

0 citationsRead paper