Toridion quantum neural networks (TQNNs) are a powerful alternative to traditional neural networks (NN) that use the principles of quantum superposition, entanglement and quantum information theory to deliver a new kind of machine learning system that can often directly replace and in some cases perform certain tasks that the classical counterpart cannot.
The similarities end generally at the name. Whereas classic NNs store information in definite structures as binary data, TQNNs use probabilistic quantum formulas to store and process the information presented to its sensory or receptive field.
Choosing a NN
Classic NNs generally come in several well known variations, the two common ones being: Convolutional neural network (CNN) and Recursive neural network (RNN). There are others such as Time delay (TDNN), Probabilistic (PNN). Although there are ways to configure and apply them to any task, from popular understanding, each type of network is suited to particular tasks. CNNs are widely thought to perform better at image processing tasks, whereas RNNs are better at processing text and speech. Because of these limitations, choosing the right kind of NN can have a huge impact on the performance of the function it underpins.
Toridion's TQNN technology addresses this by providing a NN architecture that can replace all of the above classical methodologies with a single quantum inspired system. The quantum memory architecture that underpins TQNN is both a highly advanced form of probabilistic content addressable memory together with a fault tolerant recurrent pathway system that together form powerful neuronal structures that are highly adaptive and resistant to change, whilst introducing capabilities normally reserved for biological neural systems such as Neuroplasticity and Imagination generation.
Today and the future
The availability of full scale general purpose quantum computers is said by most to be some years away, but that does not mean we cannot harness the power of quantum computational approaches and apply them to real world problems today. Toridion are are constantly working towards faster and smarter systems. By both working with key clients to evaluate real world solutions and by a constantly evolving program of research and development we are making huge strides in the fields of facial recognition, image processing, big data analytics and intelligent search systems.
I don't care how quantum machine learning works, how can I use it?
We recognise that not everyone knows or even cares how quantum works, they just want to use a better solution. To that aim we are now working to bring these powerful devices to the world in the form of a suite of simple programming APIs (in Alpha stage) that will let anyone integrate quantum machine intelligence into commercial and non commercial software solutions without having to understand the complexities of quantum computing or even Artificial Intelligence (AI) itself.
So far we have deployed TQNN in situations where it has brought performance gains of over 5600% to real world vision processing. In search engine design, TQNN has shown ability to bolt on phenomenal speed increases whilst offering an anonymous and highly secure approach to wide area data sharing.