Romain Claret

Romain Claret is a Doctoral Assistant at the Institute of Information Management within the Faculty of Economics at the University of Neuchâtel, Switzerland. He is completing his PhD in Computer Science with 85% progress, focusing on evolving neural networks that mimic human collective intelligence. His research bridges computer science, neuroscience, and cognitive science to develop next-generation artificial intelligence systems.

His research interests include:

• Evolving Neural Networks
• Neuromodulated Neural Networks
• Sparse Neural Networks
• Collective Computational Intelligence
• Knowledge representation and World models
• Neuroscience-inspired Artificial Life

Claret's research focuses on developing AI systems that evolve rather than being explicitly engineered. His work on GEENNS (Compositional Intelligence Through Evolution) demonstrates how neural networks can be taught to think in components rather than patterns, leading to more adaptable and interpretable AI. His systematic optimization approach has achieved 29% MNIST accuracy with ES-HyperNEAT, surpassing previous benchmarks through exploration of 3+ billion configurations.

His publications highlight advancements in hyperparameter optimization for evolutionary algorithms and their transferability across tasks. This research has implications for autonomous systems, personalized healthcare, and adaptive robotics where traditional AI approaches struggle with novel situations. His work proves that evolutionary approaches can yield transferable solutions between different problem domains.

Claret has received training in human subjects research, entrepreneurship, and academic writing. He is also the founder of Artificialkind, a startup focused on evolving intelligence, and serves as a Visiting Researcher at University College Dublin since September 2023.

As an educator, he has served as a Guest Lecturer at the University of Geneva and mentors students in computational thinking and evolutionary AI approaches. His teaching philosophy emphasizes that 'intelligence emerges, isn't programmed' and that 'adaptability > benchmark scores' in complex, changing environments.