"Researchers at the Max Planck Institute for the Science of Light have developed a simpler and potentially more sustainable method for implementing neural networks using optical systems." (ScitechDaily, Neural Networks Made of Light: Photon Power Drives the Next AI Revolution)
The AI requires a very secure data transmission. The purpose of ultra-secured data transmission is that the outside effects don't harm information.
When data travels in a complex structure high-level encryption guarantees that only receivers of the data can receive it. Because the entire network will not handle the same data, that leaves reserve in that network.
That guarantees that neural networks cannot be stuck. If there are some kind of problems with the photonic or electric routers the system can put information to travel in other routes.
This structure where the system follows the minimum force principle saves energy. But it allows the system to handle multiple missions at the same time. If the problem is complex, the intelligent system connects more calculation units for that problem.
The photonic networks are the new tools for computing. There are two versions of the photonic networks. The first one is the optical neural network, where electric data cables are replaced by optical cables. In that system, the information travels in the laser rays. The laser ray that is targeted to photovoltaic cells can transfer data more securely than electric systems. Eavesdropping normal electric data cables is quite easy. Eavesdroppers must just put a sensor on the cable. And observe changes in voltages.
Eavesdropping the laser systems is not so easy. To steal information from coherent light, the system must push the sensor into the laser ray. And that is seen as the decrease of the light power. If the laser ray is covered using another, hollow laser ray, that denies the eavesdropper to see that laser ray.
The intelligent system can also share information into data packages using TCP/IP protocol. And then the system can send those data packages by different routes. The system can put a serial number into each data package, and that allows the receiving system to sort them into the right order. That system can send data in mixed order, which makes it hard to open. The data packages can send in white noise or meaningless data packages, which makes attackers work harder.
Artistic illustration of a neuromorphic system of waveguides carrying light. Credit: @ CC Wanjura (ScitechDaily, Neural Networks Made of Light: Photon Power Drives the Next AI Revolution)
Quantum photonic neural network.
Quantum neural networks can revolutionize AI. In those systems, superpositioned and entangled photons form multi-state quantum computers.
A photonic neural network can be the net of the superpositioned and entangled photons trapped in the nanotubes. The photons will be trapped in the points of the carbon frames in the nanotubes.
The idea is that the carbon frames in fullerene nanotubes can stop photons at a certain distance from each other. Then that superpositioned and entangled photon series acts as a photonic neural network.
3D quantum photonic neural networks.
Theoretically is possible to make the photonic brains. The 3D quantum structure operates like quantum computers. The problem with the 3D photonic brains is this: the system requires the ability to stop photons. The trapped photons act as photonic routers and gates.
If the receiving photon's energy level is higher than the transmitter, the information cannot reach the receiver, and the gate is closed. When the transmitter sends information at a higher energy level than the receiver that energy makes the receiver oscillate and the gate is open.
The other way is to use noble gas atoms (or small fullerene molecules) to close and open the information route to the quantum nanotube lines. The idea is that those atoms are at the crossroads of those structures, and they can be used to select the route, where information can travel.
Another way is to use noble gas atoms in the structure that act as router-switch. The router can be the flap system like noble gas atoms, that close the fullerene nanotube or quantum channel. And that thing means the router is the nanotube structure where the data can travel in selected routes.
And if the system can trap photons in the photonic crystals or some kind of photonic field, that would be the greatest possible quantum computer. It's possible. Photonic crystals or artificial "industrial" diamonds will involve channels, there are 100% reflecting mirrors.
Or in the wildest visions, standing wave movement can hold photons in their position. Those waves can make photons travel in wanted directions, or even put them to hover in wanted positions.
The ability to make the 3D nanotube structure, and grow diamonds around them can make this thing possible. But that requires that the nanotube structure can keep its form in melted metal, where diamond grows. In that structure, there could be a frame at the crossroads of the nanotubes, where the photon is trapped. Information travels between those photons in the form of electromagnetic waves.
https://scitechdaily.com/neural-networks-made-of-light-photon-power-drives-the-next-ai-revolution/
Comments
Post a Comment