Latest posts tagged with #GraphNeuralNetwork on Bluesky
Neo4j’s new graph neural fraud detector nails a solid ROC but still flags every legit transaction. Curious how threshold tweaks could fix it? Dive into the numbers, confusion matrix quirks, and real‑time challenges. #Neo4j #GraphNeuralNetwork #FraudDetection
🔗 aidailypost.com/news/neo4j-g...
Railway‑centric Graph Neural Network Boosts Station Delay Forecasts
A new Railway‑centric Spatio‑Temporal Graph Convolutional Network predicts average station delays, using an Indian Railway dataset of 4,735 stations across 17 zones. getnews.me/railway-centric-graph-ne... #railway #graphneuralnetwork
Physics‑Informed GNN Improves Medium‑High Voltage Power Flow Accuracy
PIGNN‑Attn‑LS, a physics‑informed graph neural network, achieves 0.00033 pu voltage RMSE and runs 2–5× faster than Newton‑Raphson on grids up to 1,024 buses. Read more: getnews.me/physics-informed-gnn-imp... #graphneuralnetwork #powerflow
Graph‑Enhanced Vision‑Language Model for Explainable Diabetic Retinopathy Staging
Graph‑neural‑network and vision‑language model stage diabetic retinopathy from OCTA scans, improving accuracy over baseline. Read more: getnews.me/graph-enhanced-vision-la... #diabeticretinopathy #visionlanguage #graphneuralnetwork
This approach provides transparency into the classification process and could help identify new and emerging doping agents to be banned.
#cleansport #AI #XAI #ArtificialIntelligence #doping #NeuralNetworks #GraphNeuralNetwork #science #research
Imaging-based #SpatialTranscriptomics #Segmentation
Segger
👉Attention #GraphNeuralNetwork
👉Learn a transcript-cell joint latent space
👉Unassigned transcripts➡️fragments➡️
median 20 transcripts/fragment vs 50 /cell
vs Baysor BIDCell CellPose 10Xprotocol
bioRxiv 2025
www.biorxiv.org/content/10.1...
🌐"ICoN: Integration using co-attention across biological networks" introduces an unsupervised #GraphNeuralNetwork model to create robust protein representations from diverse networks.
Read more: doi.org/10.1093/bioa...
#ProteinResearch
The library consists of various dynamic and temporal geometric deep learning, embedding, and spatio-temporal regression methods from a variety of published research papers. #GNN #GraphNeuralNetwork
SpatialGlue
Spatial multi-omics integration
#GraphNeuralNetwork
Intra-modality + Inter-modality dual-attention
spatial epigenome (ATACseq, CUT&Tag)–transcriptome (Stereo-CITE-seq)–proteome (SPOTS)
vs Seurat totalVI MultiVI MOFA+ MEFISTO scMM StabMap
#NatMeth 2024
www.nature.com/articles/s41...
