New AI model enables native speakers and foreign learners to read undiacritized Arabic texts with greater fluency
Peer-Reviewed Publication
Updates every hour. Last Updated: 4-Apr-2026 03:16 ET (4-Apr-2026 07:16 GMT/UTC)
Scientists at the University of Sharjah report that they have developed a new machine-learning system designed to overcome challenges encountered in the diacritization of Arabic texts. The system mainly targets problems that existing programs face when encountering undiacritized Arabic script, writing that lacks the vowel marks necessary to pronounce words correctly, a process linguists refer to as diacritization. The presence of diacritics in Arabic is vital not only for how a word is pronounced, but also for semantics. A single word can have multiple, entirely different meanings, depending on how it is articulated.
Early cancer detection often relies on complex, invasive, and time-consuming staining procedures. A research team from Southeast University has developed a novel "label-free" biosensor that uses the physics of "band folding" to unlock high-density hidden electromagnetic modes in the sub-terahertz range. This technology creates a rich spectral fingerprint that can distinguish between normal cells and cancer cells of varying malignancy without chemical markers. By correlating macroscopic electromagnetic signals with microscopic cellular biomass density, this method offers a safe, rapid, and non-destructive path for future clinical cancer screening.
Researchers have developed a compact, high-precision gas mapping system by integrating on-chip dual microcombs with a nanomaterial-functionalized fiber sensor array. This hybrid architecture achieves the simultaneous, highly selective detection of 12 distinct gas species with a LOD (limit of detection) of 24.3 parts per billion. By leveraging the specific chemical interactions of nanomaterials driven by precise optical comb lines, the system offers a robust solution for analyzing complex gas mixtures in environmental and industrial settings.
A new study explains the brain mechanisms behind moments when we first recognize a blurry object, a primal ability that enabled our ancestors to avoid threats. Based on this understanding, the team built an AI model with a human-like perceptual mechanism that learn new tasks with little training.
The research group led by Professor Yijun Feng and Professor Ke Chen from Nanjing University reports a hybrid-phase strategy that unlocks broadband achromatic wavefront control for both circular polarizations. By combining Aharonov–Anandan and Pancharatnam–Berry geometric phases within a single-layer meta-atom, they enable independent phase and group delay design for the two spin channels, overcoming the spin-locked limitation of conventional achromatic metasurfaces. The team validates beam deflectors and metalenses in the 8–12 GHz band and presents terahertz designs for 0.8–1.2 THz, demonstrating a general dispersion-engineering route to compact, polarization-multiplexed meta-optics for broadband imaging and multi-spectral sensing. The study was published in PhotoniX on December 16, 2025.