• Unsupervised learning inspired by early development

    The ‘emergence of understanding’ is a major open challenge in the study of cognition and the brain. Vision plays a major part in this task. Current computational theories dealing with the acquisition of knowledge about the world through visual perception still cannot adequately cope effectively with natural cognitive concepts, which depend not only on statistical regularities in the sensory input, but also on their significance and meaning to the observer. As a result, current methods are inherently limited in their capacity to acquire cognitive concepts related to agents in the world, their goals and actions, social interactions and others. We aim to construct computational modeling of how knowledge of the world emerges from the combination of innate mechanisms and visual experience.


  • Aspects of visual intelligence

    A central problem in the study of vision is discovering the features and representations used by the brain, as well as the mechanisms to perceive and understand the world around us. Recent successes in computational models of visual recognition naturally raise the question: Do computer systems and the human brain use similar or different computations? In this line of research we show that the human visual system uses features and mechanisms, which are critical for recognition, but are not used by current models. We suggest directions to produce more realistic and better-performing models, in particular when current approaches have fundamental limitations.


  • Gaze perception and social intelligence

    Humans, as a social species, are remarkably adepts at understanding other people's mental states based on the perception of their actions. Developmental studies have shown that even young infants can be engaged in joint attention with other humans and can understand their mental states by observing and interpreting their looking direction. Inferring where people are looking at plays a major role in the development of communication and language, opens a window into their mental state, and serves as an important cue towards understanding intentions and actions in social interactions .


  • AI for healthcare

    Data science and the use of ML and AI methods are quickly turning into an essential and growing part of health care worldwide. Successful use of AI-based methods have been demonstrated in different application, including the automated analysis of medical imaging, such as Google’s system for predicting diabetic retinopathy progression using color fundus photographs, and there is a growing interest in including electronic health records in AI-based systems. The vision of this research is to harness the power of enormous health-related data sets worldwide from multiple sources, including medical imaging and clinical data, and use data science to improve diagnosis, suggesting therapies, predicting patient outcomes, or screening and preventing diseases.