The progress of photonics technologies such as lidar, remote sensing, and microscopy depends on better identification of light sources. Traditionally, identifying and distinguishing diverse light sources such as sunlight, laser radiation, or molecule fluorescence has required millions of measurements, especially in low-light environments, which thwarts the progress of quantum photonic technologies. Scientists have been trying for more than 20 years to reduce the number of photons and measurements required to perform imaging, remote sensing, and metrology at extremely low light levels with little progress. 

However, a team of researchers from the Department of Physics and Astronomy at Louisiana State University in the United States have recently published their work in the American Insitute of Physics’s (AIP) Applied Physics Review (APR) that demonstrates a smart quantum technology that has dramatically reduced the number of measurements required to identify light sources. “We trained an artificial neuron with the statistical fluctuations that characterize coherent and thermal light,” said Omar Magana-Loaiza, an author of the paper said in an interview featured on the AIP website. 

In order to dramatically reduce the number of measurements required to identify light sources via artificial intelligence (AI), the researchers make use of an ADALINE neuron. An acronym for the ADAptive LINear Element, ADALINE is a single neural network model based on a linear processing element that reduces the number of measurements required to discriminate signal photons from ambient photons. “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 Chenglong You, a fellow researcher and co-author on the paper.

In the training stage, ADALINE is capable of learning the correct outputs (named as output labels or classes) from a set of inputs, so-called features, by using a supervised learning algorithm. In the test stage, this neuron produces the outputs of a set of inputs that were not in the training data, taking as reference the acquired experience in the training stage. While the researchers tested ADALINE along with the naive Bayes classifier method, and both presented similarly accurate results, ADALINE requires far less computational resources than a naive Bayes classifier. “Our results indicate that a single artificial neuron outperforms the naive Bayes classifier at low light levels. Interestingly, this neuron has simple analytical and computational properties that enable low-complexity and low-cost implementations of our technique,” states the research paper.

With fewer measurements, researchers can identify light sources much more quickly, and in certain applications, such as microscopy, they can limit light damage since they don’t have to illuminate the sample nearly as many times when taking measurements. As laser light plays an important role in remote sensing, this work could also enable the development of a new family of smart Lidar systems with the capability to identify intercepted or modified information reflected from a remote object. Lidar is a remote sensing method that measures the distance to a target by illuminating the target with laser light and measuring the reflected light with a sensor.

“The probability of jamming a smart quantum lidar system will be dramatically reduced with our technology,” said Magana-Loaiza. 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.