Adrien Bardes
Papers
1
Total Citations
3
H-Index
1
About
Adrien Bardes is a researcher at the forefront of self-supervised learning and video understanding, with a particular focus on developing AI systems capable of learning rich world representations from raw, unlabeled data. His work centers on the ambitious goal of building machines that can understand, predict, and plan by observing the world — much like humans do — rather than relying on expensive labeled datasets or dense reward signals. His most notable recent contribution, **V-JEPA 2** (2025), advances this vision by combining internet-scale video data with a small amount of robotic interaction data to produce models capable of physical world understanding and planning. The work builds on the Joint Embedding Predictive Architecture (JEPA) framework, a paradigm that has garnered significant attention in the self-supervised learning community for its elegant approach to representation learning without generative reconstruction. Bardes' research sits at a critical intersection of computer vision, robotics, and cognitive-inspired AI, pushing toward models that are both data-efficient and generalizable. Though V-JEPA 2 is early in its citation trajectory with 3 citations at time of writing, its scope and ambition signal a potentially transformative contribution to how AI systems learn from video at scale.
Research Focus
Key Achievements
Top Papers
- 1