Abstract: Alessandro Oltramari

Visual Intelligence Beyond Representation
“Visual intelligence” can be conceived as the human capability to understand the relations holding within objects in a perceived scene and reason over temporal and conceptual dimensions of those relations. From philosophy to AI, this  process of understanding has been widely associated to the notion of “representation”. But is representation enough to deal with “visual intelligence”? In particular, are representations enough to create “artificial visual intelligences”? According to the mainstream, the answer is yes. In this talk I will outline the theoretical and experimental limits of the representational approach to visual intelligence, presenting insights from cognitive science.
Dr. Oltramari received his Ph.D. from University of Trento (Italy) in Cognitive Science and Education, in co-tutorship with the Institute for Cognitive Science and Technology of the Italian National Research Council (ISTC-CNR). He also has a B.S. and M.S. in Philosophy of Science (University of Padua – Italy). He has worked as a research scientist at the Laboratory for Applied Ontology (ISTC-CNR) in Trento since 2000. Alessandro has been a Visiting Research Associate at Princeton University (WordNet group) in 2006. His primary interests are in applied ontology, cognitive science, computational semantics. Most of his research questions center around how cognitive aspects of knowledge representation are involved in semantic technologies. Alessandro’s research activity at CMU mainly deals with integrating ontologies and ACT-R models. In general, extending cognitive architectures with ontologies represents a brand new area of research in Artificial General Intelligence.