Natural Neural Network (NNN)
This visualisation illustrates a natural neural network (NNN) comprising neurons whose weights are represented by continuous values. These neurons operate on and are trained by electromagnetic (photonic) sine(light)waves containing encoded, modulated data. Unlike conventional digital systems based on binary (0 or 1) square waves, this approach exhibits greater variability and more closely resembles natural processes.
It is assumed that matrix multiplication of this NNN will still be possible using a photonic computer system.
The Behaviour of Neurons within the Natural Neural Network (NNN)
The neurons within the Natural Neural Network (NNN) initially produce a linear response as their output. At this stage, the relationship between the input and output is straightforward, with the output directly reflecting the input in a proportional manner.
Transition to Non-Linear Output Profiles
Upon training, the output profile of these neurons becomes far more nuanced. The process of training introduces non-linearity, which results in the amplification or attenuation of the output response. This means that the neurons no longer simply mirror the input; instead, they are capable of adjusting their responses, either enhancing or suppressing the output as required.
Mixing of Input Signals
As the non-linear profile is learnt, the mixing of input signals is also influenced. The neurons integrate the various inputs in a manner that is shaped by the training process, allowing for more complex and sophisticated responses that go far beyond the initial linear behaviour.
ElectroMagnetic Wave Logic (EMWL)
ElectroMagnetic Wave Logic (EMWL) is promising for next-generation AI deep learning networks; as it offers scalability without losing real-world resolution by inputting directly upconverted (transverted) real-sound, real-radio and real-vision EM waves into the neural network forgoing the need for lossy ADC processes.
Quantum Photonics and Logic Gates
These illustrations depict a set of waves with continuous values (rather than the traditional binary pairing of 0 or 1) entering logic gates. The gate combines the two input signals to produce a continuous output rather than binary that ranges from TRUE (Positive) to FALSE (Negative)
Continuous Value AND Gate 
Continuous Value OR Gate 
Continuous Value XOR Gate 
Quantum Entanglement of Neural Networks
A fun way to explain the concept of Entangled Quantum Neural Networks.
Quantum Entanglement: A high energy photon can be split into lower energy photons that are synchronised (SPDC) together. Imagine a big chocolate egg that is melted down into smaller chocolate eggs which then spin and rotate in lockstep
Quantum Superposition: A photon can exist at infinite fractional values between 0 and 1 at the same time and you only pick a value once it’s observed (so, view an egg!)
Quantum Entanglement of Neural Networks
This construct forms the basis for developing a Quantum Neural Network (QNN), envisioned to carry infinite numbers (weights) in parallel with infinite hidden layers, greatly expanding the potential computational power and complexity of traditional neural network architectures.