Browse Articles

A reconfigurable photosensitive split-floating-gate memory for neuromorphic computing and nonlinear activation

Abstract The rapid growth of artificial intelligence and the Internet of Things calls for compact hardware platforms that integrate sensing, computing, and nonlinear processing within a unified architecture. However, most existing neuromorphic systems implement only partial functionalities and rely on heterogeneous device integration, limiting scalability and efficiency. Here, we show a high-speed, reconfigurable multi-modal split-floating-gate memory that monolithically integrates in-sensor computing, in-memory computing, and multiple nonlinear activation functions within a single device structure. By programming charges in spatially separated floating gates, the device enables non-volatile analog control of photoresponsivity and conductance, as well as electrically reconfigurable rectification to emulate ReLU and Sigmoid activations. We further demonstrate a fully hardware-implemented sensor–processor system based on the multi-modal split-floating-gate memory arrays that performs complete unsupervised and supervised learning tasks. This work establishes a compact, energy-efficient, and reconfigurable hardware foundation for scalable intelligent systems beyond conventional silicon architectures.

Correction: Correlation analysis between RAS gene mutations and pathological morphological features in colorectal cancer

Conserved Transmembrane Asparagine Is Essential for the Ion-Conducting Structure and Dynamics of the SARS-CoV-2 Envelope Protein

Decoding the gradient-distributed colour centers in electrochromic WO3

Self-supervised learning on graphs predicts non-coding RNA and disease associations

Abstract Non-coding RNAs (ncRNAs) play crucial roles in regulating the initiation and progression of various cancers. Accurate identification disease-related ncRNAs would provide a unique opportunity to design better therapeutic interventions. Graph convolutional network-based methods have been proposed to identify potential ncRNA-disease associations (RDAs). However, some methods only use the graph structure and ignore the similarity information of nodes, and some methods integrate multi-source relation data which will introduce noise and have poor generalization. Learning robust node embeddings using graph convolutional network to build RDA predictive frameworks with high generalization remains a key challenge. We proposed a new RDA prediction scheme, SSLGRDA, composed of graph self-supervised learning and machine learning. Since SSLGRDA works on both heterogeneous and homogeneous graphs, we constructed ncRNA-disease heterogeneous graph based on known RDA, as well as heterogeneous graph based on known RDA, ncRNA similarities and disease similarities. For the latter, we ignored the node and edge types and treated it as homogeneous graph. Based on multiple contrastive or generate strategies, we used graph self-supervised learning to extract robust ncRNA and disease embedding to enhance the prediction ability and generalization of the model. Finally, we use machine learning methods to predict latent RDA probabilities. To evaluate model performance, we performed SSLGRDA on 9 ncRNA-disease datasets. Comprehensive experimental results show that SSLGRDA not only has good generalization, but also outperforms several state-of-the-art methods. Case studies on three ncRNA-disease datasets further demonstrate the ability of SSLGRDA in discovering potential ncRNA-disease associations.

Interpreting X-ray Diffraction Patterns of Metal–Organic Frameworks via Generative Artificial Intelligence

Author Correction: Factor XII signaling via uPAR-integrin β1 axis promotes tubular senescence in diabetic kidney disease

Impact of the war on forest ecosystem in Ukraine based on Sentinel-2 data

Abstract Forests play a vital role in ecology, economy, urban planning, and social well-being, emphasising the importance of monitoring forest cover and its changes. The study evaluates the ecological impact of the military conflict on forest ecosystem in Ukraine using a time series of Sentinel-2 data and machine learning algorithms. Forest losses following the beginning of the war were derived using a change detection method across the study areas. Two types of forest loss were delineated: conversion of woody areas to non-woody cover, and burnt forest. Before the war, forest loss was predominantly due to the conversion of woody to non-woody cover, accounting for 74% of total changes, while forest fires represented the remaining 26%. Following the outbreak of the conflict, the total area of forest loss doubled. Notably, the proportion of forest converted to non-woody cover decreased to 66%, while the proportion of burnt forest increased to 34%, evidencing the severe impact of military operations on forest ecosystem. Of interest, the area of forest converted to non-wood cover doubled between, possibly reflecting an increased demand for wood due to the conflict, potentially driven by a rise in legal and illegal logging as a result of weakened governance and reduced enforcement of environmental regulations.

Deep Learning Guided Exploration of Transition Metal Oxide Catalysts in Acetylene Selective Hydrogenation

Aged skin exacerbates experimental osteoarthritis via enhanced IL-36R signaling

Diagnostic performance evaluation of Elecsys anti-SARS-CoV-2 assay against RT-PCR for SARS-CoV-2 detection and surveillance in Ethiopian referral hospitals: cross-sectional study

What Determines the Directionality of Kinesins? Studying the Anomalous Directionality of Nonclaret Disjunctional Motor Mutants

Variations in the Natural History of High-Risk HPV Types Following HPV-16/18 Bivalent Vaccination in Females Aged 18-45 Years

Surface pretreatments and erosive aging effects on the bond strength of CAD/CAM resin-based materials with a self-adhesive resin cement

Ortho–Ortho Selective Oxidative Coupling of Phenols by Hydroxo Multicopper(II) Clusters

Author Correction: Revisiting the soil carbon saturation concept to inform a risk index in European agricultural soils

Short-term effects of haze exposure on clinical and subclinical cardiovascular indicators among adults in Nanjing

Fine-Tuning of the Neuropeptide Y1 G Protein-Coupled Receptor by the Tryptophan ^6.48 “Toggle Switch”

A scalable two-step genome editing strategy for generating full-length gene-humanized mice at diverse genomic loci

Metabolic interactions between bacterial co-isolates from catheter-associated urinary tract infections

Abstract Catheter-associated urinary tract infections (CAUTI) are complex infections often involving multi-species bacteria. Escherichia coli is frequently an early coloniser. Subsequent colonisation by Pseudomonas aeruginosa and coexistence mechanisms between the two strains within urethral catheters is not yet fully understood. In this study, metabolic adaptations between co-isolated clinical E. coli and P. aeruginosa strains were investigated. It was found that P. aeruginosa outgrew E. coli in artificial urine medium (AUM), whereas E. coli dominated in culture broth such as Iso-sensitest. No evidence of direct antagonism was observed. Metabolite analyses revealed distinct metabolite patterns indicating cross-feeding and metabolic adaptations. In AUM, stress-response metabolites were elevated. Additionally, E. coli appeared to experience Fe-limitation in AUM, while the same was not observed for P. aeruginosa. The results highlight the influence of nutrient conditions on processes within mixed biofilms.