Learning algorithms for probabilistic neural nets

Although neural net models show promise in areas where traditional AI approaches falter, such as pattern recognition, pattern completion and content addressable memory, their success is constrained by slow learning rates and the difficulty of physical implementation; learning strategies such as error-back-propagation are also implausible as biological models. The Probabilistic Logic Neuron (PLN) represents an attempt to address these issues while retaining the emergent properties of the traditional connectionist models. It is implemented as a variable logic device, with the consequences that it can perform any Boolean function of its inputs, while its use in nets allows drastic reduction in quantity and specificity of connection requirements.