PulseAugur / Brief
EN
LIVE 14:28:36

Brief

last 24h
[2/2] 222 sources

Multi-source AI news clustered, deduplicated, and scored 0–100 across authority, cluster strength, headline signal, and time decay.

  1. RESEARCH · arXiv cs.AI English(EN) · · [2 sources]

    GC-MoE: Genomics-Guided Cell-Type-Specific Mixture of Experts for Histology-Based Single-Cell Spatial Transcriptomics

    Researchers have developed GC-MoE, a novel method for estimating gene expression in individual cells from histopathological images. This approach utilizes a Mixture-of-Experts model guided by genomics to predict cell-type probabilities and gene expression, aiming to reduce the need for expensive single-cell measurements. The system incorporates cell-type-specific predictors and attention modules to capture gene programs and neighboring cell context, showing improved results over existing methods. AI

    IMPACT This method could significantly reduce the cost and complexity of spatial transcriptomics research by enabling gene expression prediction from standard histology images.

  2. TOOL · arXiv cs.AI English(EN) ·

    Graph-Conditioned Mixture of Graph Neural Network Experts for Traffic Forecasting

    Researchers have developed a new framework called GC-MoE for traffic forecasting that utilizes a mixture of graph neural network experts. This approach allows for personalized combinations of frozen forecasting experts for each node, adapting to local graph topology and recent traffic data. The system trains only a small routing module while leveraging pre-trained experts, showing improved Mean Absolute Error on standard benchmarks. AI

    IMPACT This specialized GNN approach could improve the accuracy of real-time traffic prediction systems.