Information storage in high-order associative memory with unequal neural activity

Summary form only given. Neural networks with high-order interactions provide satisfactory stored patterns and error corrections. Such interactions increase the storage capacity of the networks and allow a class of problems to be solved which were intractable with standard networks. An analysis is made of the capacity of these higher order networks by statistical methods. It is shown that the probability of the states of neurons being active and passive can always be chosen with a probability of 0.5 each.<>