Interpretable deep learning network significantly improves tropical cyclone intensity forecast accuracy

Researchers explore the human immune system by looking at the active components, namely the various genes and cells involved. But there is a broad range of these, and observations necessarily produce vast amounts of data. For the first time, researchers including those from the University of Tokyo built a software tool which leverages artificial intelligence to not only offer a more consistent analysis of these cells at speed but also categorizes them and aims to spot novel patterns people have not yet seen.

Organic photovoltaics (OPVs) have achieved remarkable progress, with laboratory-scale single-junction devices now demonstrating power conversion efficiencies (PCEs) exceeding 20%. However, these efficiencies are highly dependent on the thickness of the photoactive layer, which is typically around 100 nm. This sensitivity poses a challenge for industrial-scale fabrication. Achieving high PCEs in thick-film OPVs is therefore essential. This review systematically examines recent advancements in thick-film OPVs, focusing on the fundamental mechanisms that lead to efficiency loss and strategies to enhance performance. We provide a comprehensive analysis spanning the complete photovoltaic process chain: from initial exciton generation and diffusion dynamics, through dissociation mechanisms, to subsequent charge-carrier transport, balance optimization, and final collection efficiency. Particular emphasis is placed on cutting-edge solutions in molecular engineering and device architecture optimization. By synthesizing these interdisciplinary approaches and investigating the potential contributions in stability, cost, and machine learning aspects, this work establishes comprehensive guidelines for designing high-performance OPVs devices with minimal thickness dependence, ultimately aiming to bridge the gap between laboratory achievements and industrial manufacturing requirements.

Identifying and interpreting vacancies in patent maps is a promising approach to discover technological opportunities. However, it remains a challenging task. Recently, scientists from Seoul National University of Science and Technology have developed an innovative machine learning approach based on text-embedding inversion, which translates patent vacancies into human-readable formats, helping to uncover technological opportunities for corporate growth.

A research team from the Institute of Physics, Chinese Academy of Sciences, has developed FastTrack, a new machine learning-based framework dedicated  to evaluate ion migration barriers in crystalline solids. By combining machine learning force field (MLFFs) with three-dimensional potential energy surface (PES) sampling and interpolation, FastTrack enables accurate prediction of atomic migration barriers within mere minutes. Unlike traditional methods such as density functional theory (DFT) and nudged elastic band (NEB), which can take hours or days per calculation. FastTrack offers a speedup of over 100 times without sacrificing accuracy, closely matching experimental and quantum-mechanical benchmarks. This powerful tool automatically identifies diffusion pathways, visualizes energy landscapes, and provides detailed microscopic insights into ion migration mechanisms, crucial for designing more efficient batteries, fuel cells, and other energy storage and conversion devices.

SUTD researchers have developed a streamlined life cycle assessment method that makes environmental evaluation faster, cheaper, and more accessible to product designers without compromising reliability.