Raia Hadsell is an American research scientist. She is the Vice President of Research at Google DeepMind, where she oversees research in robotics, neural network safety, and navigation.
Contrastive Learning at NYU
Hadsell completed her Ph.D. at New York University in 2008, where her advisor was Yann LeCun. Her doctoral research pioneered **contrastive learning** (introducing the contrastive loss function), which allows neural networks to learn representations by comparing positive and negative pairs of inputs. Contrastive learning is a key technique in modern computer vision and self-supervised learning (like CLIP).
DeepMind: Progressive Networks and Navigation
At DeepMind, Hadsell pioneered **progressive neural networks**, which allow AI agents to learn new tasks sequentially while preventing "catastrophic forgetting." She also led the robotics team that developed deep reinforcement learning agents that can navigate real-world environments like streets and mazes without maps.
