A group of researchers has used artificial intelligence (AI) in order to identify light sources. The new method requires drastically fewer measurements than what is traditionally required.
Many photonic technologies including lidar, remote sensing, and microscopy are developed in part by identifying sources of light. Some of these sources include sunlight, laser radiation, and molecule fluorescence. Identifying them normally requires millions of measurements, which is especially true in low-light environments, making it extremely difficult to implement quantum photonic technologies.
The work was published in Applied Physics Reviews, from AIP Publishing. It is titled “Identification of light sources using machine learning.”
Omar Magana-Loaiza is an author of the paper.
“We trained an artificial neuron with the statistical fluctuations that characterize coherent and thermal light,” said Magana-Loaiza.
The artificial neuron was first trained with light sources, which led to it being capable of identifying certain features that are associated with specific types of light.
Chenglong You is a fellow researcher and co-author of the paper.
“A single neuron is enough to dramatically reduce the number of measurements needed to identify a light source from millions to less than hundred,” said You.
Applications and Benefits
Because there are such fewer measurements required in order to identify light sources, it can be done much faster. Besides being faster, there can be a reduction in light damage. For example, light damage can be limited in microscopy since the sample does not need to be illuminated as much as when many measurements are required.
Roberto de J. León-Montiel is another co-author of the paper.
“If you were doing an imaging experiment with delicate fluorescent molecular complexes, for example, you could reduce the time the sample is exposed to light and minimize any photodamage,” said León-Montiel.
Another area that will benefit from this technology is cryptography, where millions of measurements are often required to generate keys to encrypt messages or emails.
“We could speed up the generation of quantum keys for encryption using a similar neuron,” said Magana-Loaiza.
Laser light, which is important in remote sensing, could also benefit. A new family of smart lidar systems could be developed, capable of identifying intercepted or modified data that is reflected from a remote object. Lidar is a remote sensing method that illuminates a target with laser light. It then measures the reflected light with a sensor in order to measure the distance to a target.
“The probability of jamming a smart quantum lidar system will be dramatically reduced with our technology,” Magana-Loaiza continued. In addition, the possibility to discriminate lidar photons from environmental light such as sunlight will have important implications for remote sensing at low-light levels.
- Nanoantenna Enables Advanced Quantum Communication and Data Storage
- Human Brain Project Releases New Paper on Exascale Computing Power
- AI Researchers Estimate 97% Of EU Websites Fail GDPR Privacy Requirements- Especially User Profiling
- 5 Best Machine Learning & AI Books of All Time
- Earkick Raises $1M for Real-Time Mental Health Tracker