Shirin Dora
Papers
2
Total Citations
23
H-Index
2
About
Shirin Dora’s research lies at the exciting intersection of cognitive neuroscience and robotics, where she develops biologically plausible learning algorithms for autonomous systems. Her primary focus is on multimodal representation learning, particularly how machines can integrate vision and touch—much like the human brain—to robustly recognize places and navigate complex environments. Dora’s most notable contribution is the introduction of deep Hebbian predictive coding for place recognition, a framework that allows robots to bind information from different sensory sources without requiring labeled data. Her 2021 paper on this topic, which has garnered 21 citations, demonstrates how predictive coding principles can overcome challenges like data registration mismatches, making robot navigation more resilient. In earlier work, she proposed MuPNet (Multi-modal Predictive Coding Network), a biologically inspired architecture that extracts joint visuo-tactile latent representations through unsupervised learning. Though a niche contribution, this work lays essential groundwork for embodied AI systems that learn from touch and sight simultaneously. Dora’s research is particularly impactful for students and engineers interested in neuromorphic computing, sensor fusion, and creating robots that perceive the world as seamlessly as living organisms do.
Research Focus
Key Achievements
Top Papers
- 1
- 2