I am a PhD student in the Computer Vision lab at the Weizmann Institute of Science.
My Advisor is Prof. Shimon Ullman.
My Current Research
I work on the Digital Baby project.
Our goal is to teach computers to perceive visual input by imitating the way babies learn visual perception.
We study how babies learn to see and apply similar learning rules to computer systems.
This approach of cognitive learning has two key advantages over standard supervised learning:
- Learn from unlabeled videos. No need for manually labeled data.
- Learning is not limited by existing training data. Visual recognition can continue to develop as more visual input is received.
As research progresses, we also gain insights into human learning.
In order to learn useful visual concepts, one has to focus on subjectively interesting data within the entire visual stimuli.
We use various cognitive cues and learning rules in order to achieve this. In the process we discover interesting learning principals,
which are useful for computers and are likely to be used by humans as well.
Research Interests
- Computer and Human Vision
- Object Recognition and Scene Understanding
- Statistical Machine Learning
- Cognitive Learning (Unupervised Learning based on Cognitive Principals)
- Human Cognitive Development