Quantum neural network
Quantum neural networks (QNN)refers to the class of neural network models, artificial or biological, wich rely on principals inspired one or way or another from quantum mechanics in general, usually more specifically inspired from quantum computing.
Three diffrent classes can be generally distinguished:
1- The quantum neural network proposed by Purushothaman & Karayiannis in 1997: This QNN si only implicitly linked to quantum mechanics, and is not very different from conventional feedforward networks. It's quantum aspect comes from the fact that it uses a superposition of sigmoids instead of just one sigmoid as an activation function in each neuron. This gives the network the ability to perform inherent fuzzy classification, as opposed to other neuro-fuzzy schemes which require the degree of fuzziness to be learned or be present as an input.
This is also currently the QNN model which is used the most in practical applications (in terms of published research papers).
2- The class of quantum neural networks which explicitly use concepts from quantum computing, such as superposition, interference, entanglment or qubits and qubit registers. Several authors have published papers on this type of QNN, however most have remained at the purely theoretical level, especially since most proposals require a functional quantum computer to be implemented.
3- Models of biological neural networks (e.g animal and human brains) which use concepts from quantum computing and quantum mechanics to explain the exceptional performance of biological brain as opposed to conventional computing devices, or to explain why humans (and eventually other animals) exhibit consciousness, while current computers do not.