ANITI is delighted to invite you to an exceptional scientific seminar, held on the occasion of the visit to Toulouse of two internationally renowned researchers.
📅 Thursday, July 17, 2025
🕥 Starting at 10:30 AM
📍 B612 — Room 6, Toulouse
🍽️ Followed by a lunch buffet
——
Thomas Serre (Brown University & ANITI)
Scaling Laws vs. Neural Laws: Toward More Natural Artificial Vision
While large vision models now match human accuracy on benchmarks like ImageNet, their visual strategies diverge profoundly from those of the brain — and this divergence, strikingly, widens with scale. Thomas Serre will argue for an approach to artificial vision grounded in the neural laws of biological vision: developmental principles, cortical inductive biases, and learning from naturalistic video. Recent results from his lab show how this approach can pull deep networks toward markedly more human-like visual strategies, and how brain-inspired alternatives to transformer self-attention open promising new directions.
Ludovic Righetti (New York University & ANITI)
Learning Robotic Behaviors with Optimal Control and World-Models
How can planning and control algorithms operate directly in sensor space — vision, touch — rather than in abstract model-based representations? Ludovic Righetti will present his recent work on learning world-models and designing reinforcement learning algorithms tailored to multimodal sensory data, with concrete applications to legged locomotion and manipulation. The talk will also open onto a broader discussion on the societal impacts of robotics research and the responsibilities that engineers and researchers should embrace.
——
Both talks speak directly to the core themes of ANITI: trustworthy artificial vision, learning-based robotics, and the dialogue between cognitive science and artificial intelligence.
The seminar will be followed by a lunch buffet, offering a privileged opportunity to exchange with both speakers.
👉 Attendance is free but registration is required
English 
French