The most commonly used model for Artificial Neural Networks (ANNs) is based on neurons that implement their inherent functionality by computing a weighted sum of the provided inputs and yielding their result by applying an activation function. The challenge in this model is to infer the right set of weights that should be used to ensure learning capacity. The training process is usually time-consuming and error-prone. An alternative model is the RAM-based neural network, wherein a neuron is viewed as a simple look-up-table addressed by the input to access the stored data from which an output is derived.
Weightless neural systems
De Gregorio M;
2016
Abstract
The most commonly used model for Artificial Neural Networks (ANNs) is based on neurons that implement their inherent functionality by computing a weighted sum of the provided inputs and yielding their result by applying an activation function. The challenge in this model is to infer the right set of weights that should be used to ensure learning capacity. The training process is usually time-consuming and error-prone. An alternative model is the RAM-based neural network, wherein a neuron is viewed as a simple look-up-table addressed by the input to access the stored data from which an output is derived.File in questo prodotto:
Non ci sono file associati a questo prodotto.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
