The rapid rise of IoT and AI is exposing the energy and data movement limits of von Neumann computing. Neuromorphic computing, especially spiking neural networks, offers a compelling alternative but demands compact hardware with neuron like nonlinear dynamics. In the new Nano Research review, Prof. Chuanyu Han’s team highlights recent progress in artificial spiking neurons built from emerging electronic devices, spanning memristor-based and transistor-based implementations. The article distills key spiking mechanisms, practical circuit design templates, and system demonstrations for neuromorphic perception and brain inspired computing, and introduces a rigorous unified benchmarking framework to help compare technologies and accelerate interdisciplinary work across materials science and microelectronics.

SJTU researchers propose a dynamic MCPV model and CZone mechanism, slashing memory footprint by 90%.

In a paper published in SCIENCE CHINA Earth Sciences, a team led by Professor Chuanfeng Zhao from Peking University presents a comprehensive review of global sea-land breeze research. The study systematically synthesizes recent advances in identification methods, key observational benchmarks, and the complex meteorological and non-meteorological factors that govern these coastal wind systems. The authors highlight that future breakthroughs will depend on improving spatiotemporal precision and clarifying how aerosols influence sea-land breeze dynamics through micro-physical mechanisms.

The high annotation costs in medical detection tasks make semi-supervised object detection (SSOD) methods particularly promising. However, current research has paid little attention to medical image SSOD, and conventional SSOD approaches struggle with the unique challenges of medical imaging—scarce annotations and difficult-to-learn features.

Sequential recommendation is an important research task in the field of recommendation systems. The goal of the sequential recommendation task can be defined as follows: given a user’s interaction sequence and the associated timestamp sequence, predict the next item that the user is likely to interact with at the next time step. The existing methods have limitations in capturing the diversity of long-term interests, the dynamics of short-term user interest, and the hierarchical relationship between them.

Sub-headline: Zhejiang University researchers introduce K-SELinux, leveraging the K framework to automate the verification of complex security policies