Cognitive Diagnosis (CD) plays a crucial role in personalized learning by evaluating students' mastery of various concepts. However, current CD models face a significant challenge—the "student-concept sparsity barrier." This occurs when students have limited interactions with certain concepts.

Existing swirling combustion technology, which relies on faulty coal, is unable to meet deep peak shaving demands without auxiliary methods. This paper developed a deep peak regulation burner (DPRB) to achieve stable combustion at 15%–30% of the boiler’s rated load without auxiliary support. Gas-particle tests, industrial trials, and transient numerical simulations were conducted to evaluate the burner’s performance. At full rated load, the DPRB formed a central recirculation zone (RZ) with a length of 1.5d and a diameter of 0.58d (where d represents the outlet diameter). At 40%, 20%, and 15% rated loads, the RZ became annular, with diameters of 0.30d, 0.40d, and 0.39d, respectively, with a length of 1.0d. At 20% and 15% rated loads, the recirculation peak and the range of particle volume flux were comparable to those at 40% rated load. The prototype burner demonstrated that, without oil support, the gas temperature within 0 to 1.8 m from the primary air outlet remained below 609 °C, insufficient to ignite faulty coal. As the load rate increased from 20% to 30%, the prototype’s central region temperature remained low, with a maximum of 750 °C between 0 and 2.0 m. In contrast, the DPRB’s central region temperature reached 750 °C at around 0.65–0.70 m. At a 3%·min−1 load-up rate, when the load increased from 20% to 30%, the prototype burner extinguished after 30 s. However, the DPRB maintained stable combustion throughout the process.

Adversarial examples—images subtly altered to mislead AI systems—are used to test the reliability of deep neural networks. However, existing methods often produce images with unnatural noise that is easy to detect. In a recent study, researchers from Japan developed “IFAP,” a new framework that aligns adversarial noise with the spectral characteristics of the original image. Extensive tests show that IFAP generates more natural-looking perturbations while remaining highly effective and resistant to common defenses.

How do cutting-edge science and technology respond to ethical and legal issues when incorporated into society? These issues are known as ethical, legal and social issues, or “ELSI” for short, and research on these issues is being carried out both within Japan and around the world. In 2023, Kobe University launched KOBELSI, its university-wide research project on ELSI in life sciences and natural sciences. As revolutionary technology in fields such as medicine and artificial intelligence (AI) continue to pop up one after another, how has ELSI research advanced? We spoke with Professor CHATANI Naoto of the Graduate School of Humanities, leader of the project and an expert in ancient Greek philosophy and bioethics, about the current state of ELSI research and its future prospects.

PPPL is leading STELLAR-AI, a new platform that pairs artificial intelligence with high performance computing to dramatically speed fusion energy simulations and connect computing resources directly to experimental devices like NSTX-U. The project, part of the national Genesis Mission, brings together national labs, universities, tech giants and fusion companies to build the computational foundation the fusion community needs.

Using biofeedback and gamification, the solution teaches emotional self-regulation strategies to child patients with disorders such as ADHD and ASD.

Generative AI is reshaping software development – and fast. A new study published in Science shows that AI-assisted coding is spreading rapidly, though unevenly: in the U.S., the share of new code relying on AI rose from 5% in 2022 to 29% in early 2025, compared with just 12% in China. AI usage is highest among less experienced programmers, but productivity gains go to seasoned developers.