| Resumen |
Hyperdimensional computing is an emergent model of computation based on the manipulation of high-dimensional vectors which are used not only to represent variables and values, but also to represent complex structures such as relations, sets and sequences. All vectors in the model are always the same size either if they represent a single concept or a sequence of objects. Hyperdimensional computing uses reduced representations, since there is a compression process to encode complex structures while maintaining the same size on the output vector. In this paper we explore the storing capacity of hyperdimensional vectors that encode semantic feature norms. We describe a method for encoding and retrieving feature information of concrete concepts and present experimental results of the successful retrieval of such features. |
