Data-driven computational models of exemplar and prototype representations

Mental representations of entities and concepts allow us to navigate and interact with the world, to retain information and use it when needed. As encodings of incoming information, mental representations are to a large extent formed and influenced by our experience with the world. Using large naturalistic databases approximating this experience and data-driven computational models to learn from it, we can estimate mental representations for individual exemplars as well as concept prototypes in a precise, quantitative manner. In this talk, I am going to present two prominent classes of such models: On the one hand, distributional semantic models or language models learn their representations from language corpora as a proxy for language experience, and on the other hand, deep convolutional neural networks learn their representations from image collections as a proxy for visual experience. This talk will outline how these models learn their representations, and present evaluations of these models from a cognitive perspective.

Hybrid Talk: https://arc.mpib-berlin.mpg.de/public-with-recording/

Meeting ID: 617 4003 6343

Passcode: 337436