sparse-distributed-memory

This project is a simulator for Sparse Distributed Memory. It is implemented using matlab and based on the paper by P. Kanerva.

Following is the description of useful functions in the SDM simluator

sdm(L, N, M, hammingRadius, bucketSize)

L is the number of hard-locations in the SDM

N is the number of bits in the location address

M is the number of bits in the memory vector

hammingRadius is defined for read and write operations that access hard-address based on hamming distance between hard-addresses and provided address

bucket-size defines the higher and lower threshold of single memory buckets

write(obj, vector, vector_address)

this function is used to write the value 'vector' at location-addresses that is within hammingRadius of provided vector_address

write-associative(obj, vector)

this function is used to write 'vector' into at location-addresses that is within hammingRadius of vector. For using this function, the dimension of location addresses and memory-vectors should be equal.

read(obj, vector_address)

this function is used to read vector form the memory such that the product vector is a combination of vector from address-locations that are within hammingRadius of vector_address

read_associative(obj, vector_address, itr)

this function is used to read from memory when the dimension of location addresses and memory-vectors are equal. Based on the provided vector_address and itr, the return value should converge at hard-address location.