[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Use large, orthogonal vectors



Our vectors are quite small, and should also have each of their elements
initialised (pseudo)randomly. Currently we're activating one feature,
but that's a local representation.

Maybe use floats, initialised with a normal distribution as per:

Cox, G., Kachergis, G., Recchia, G., & Jones, M. N. (2011).
Towards a scalable holographic word-form representation.
Behavior Research Methods, 43, 602-615.