MIT is using Artificial Intelligence (AI) for development of specialized Nanoparticles

0
712
Photo Credit: Pixabay.com

Specialized nanoparticles can be used for a variety of cloaking techniques as well as reflective surfaces. The development of these nanoparticles has been used from everything from troop operations to providing light scattering properties that can reduce heat and control design in our world. MIT researchers are now using computer modeling to design particles more thoroughly and improve their light scattering behaviors.

Multilayered nanoparticles could be of use in biomedical devices, future displays and cloaking systems within the future. Using neural networks to examine the structure of nanoparticles and then simulate the way that it scatters along different colors of light can cut down on the expensive development and improve efficiency. Specialized nanoparticles are currently tested using more practical methods, but by generating thousands of testing a computer model, it’s possible to generate an inverse design based on the findings of the protocol.

This approach could one day lead to better physical applications with nanoparticles. As the machine learning solutions can effectively perform thousands of tests in intensive simulations possible to tackle a series of problems with testing and rollout using these neural networks.

As the system improves, it could be possible for these computer systems to one day perform intelligent testing that will improve almost any type of physical display system utilizing this nanotechnology. The Understanding of a new technique that could be suitable and which exact specification would work best for a new product can be done at a rapid pace.

The problem with the development in current nanoparticles is that almost every particle is layered together like an onion. Every single layer within the material is made of something, and this can cause them to scatter in different wavelengths of incoming beams of light. Calculating the effects of every layer of nanoparticles often involves intensive computational study. As the complexity of the nanoparticle increases the complexity of the simulation also increases alongside it.

The simulations are going even further in predicting new ways that particles can scatter light. Using known examples as the baseline, the program is creating simulations using new combinations and generating patterns in which the neural network can extrapolate findings. Simulations are exact, and they can be reproduced thousands of times to eventually create patterns on a graph of which types of particles can scatter specific types of wavelengths in light. The accuracy currently isn’t on an exact par with a physical test, but it is much faster and repeatable than a physical simulation.

Future training of the network will involve going back and running the program in reverse so that the program might eventually be able to decipher the exact combination to achieve a specific wavelength output. This reverse engineering design could one day be the authority in nanoparticles.

At its current stage of development, the program is aiding scientists in cutting development time on their newest Nanoparticle projects from a series of days down to just a few minutes in some cases. As technology improves and can one day deliver answers on nanoparticle layer combinations, it could shape a new world of display/cloaking technology.


Source: http://news.mit.edu/2018/ai-based-method-could-speed-development-specialized-nanoparticles-0601

LEAVE A REPLY

Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.