Chinese scientists proposed an anti-interference diffractive deep neural network (AI D²NN) designed for object recognition in multi-object scenarios. Unlike conventional optical neural networks built for single-object classification, AI D²NN achieves robust target recognition while effectively suppressing interference from other objects. By incorporating optical multi-dimensional multiplexing, the system enables flexible, high-capacity parallel recognition of multiple targets. This approach is expected to accelerate the practical deployment of optical neural networks in autonomous driving, medical diagnostics, and security monitoring.

Using machine learning regression models, we predict porosity (measure of potential storage volume) and permeability (measure of fluid flow property in porous media) based on petrographic thin section analysis. Histogram-based Gradient Boosting algorithms successfully predict porosity (RMSE=2.23%) and permeability (RMSE=0.64 (orders of magnitude)) based on textural mineralogical properties recorded in petrographic samples. This may in the future allow to reduce the amount of resource- and time-intensive retrieval of core material in certain use cases, as the same methods can be applied to cuttings, produced as a by-product of any drilling operation worldwide.

Optical neural networks (ONNs) offer a pathway toward low-latency, energy-efficient artificial intelligence (AI); however, their scalability in terms of parameter count remains constrained. Addressing this challenge, a research team from The Chinese University of Hong Kong has developed a metasurface-based optical learning machine that integrates 41 million parameters, achieving performance comparable to state-of-the-art AI models. This approach experimentally enables highly scalable machine vision, thereby offering a practical pathway toward large-scale, high-performance optical AI computing.

For the first time, an international research team has harnessed artificial intelligence (AI) to decode the rules of an ancient board game, pioneering a new way to reveal long-lost historical secrets.Flinders University computer scientist, Dr Matthew Stephenson, says that using modern AI techniques can bridge the gap between historical and computational studies of games.

In a survey study of U.S. teens, more than half (55.3%) reported that they had created at least one image using nudification tools, which use generative artificial intelligence (GenAI) to show what an individual may look like without clothing. Chad Steel of George Mason University, Virginia, U.S., presents these findings in the open-access journal PLOS One on March 18, 2026.