SEMINAR

Efficient Retrieval from Sparse Associative Memory

Best-match retrieval of data from memory which is sparse in feature space (i.e., hundreds or thousands of features) is a time-consuming process for sequential machines. However, a connectionist network used as a hashing function can allow faster-than-linear probabilistic retrieval from such memory, even when presented with probing feature vectors which are noisy or partially specified. I shall discuss the basic Connectionist-Hashed Associative Memory and its modifications, give theoretical results applicable to storage/retrieval of memory items represented by feature vectors made up of up to 1000 randomly selected bivalent components, and experimental results on correlated feature vectors in a spelling correction application.

WHERE: TEC 340

WHEN(day): Friday, May 1st, 1998

WHEN(time): 2:00 - 2:50 PM

EVERYBODY IS INVITED