Tag: contrastive learning

A new framework combining a 120,000-entity knowledge graph with contrastive learning outperforms traditional and graph-based baselines in evaluating digital economy ...

Researchers have introduced HOFA, a contrastive learning framework that mines triangular motifs in social networks and aligns embedding features to ...

A new semantic-guided graph attention framework improves rating prediction and ranking on Amazon cross-domain recommendation pairs while showing controlled but ...

A dual-stream graph attention network trained on conserved gut microbial network structures achieved an ROC-AUC of 0.9223 for type 2 ...

A new contrastive learning method called MsRCL uses multi-scale patch decomposition and channel-aware attention to improve anomaly detection in multivariate ...

A new contrastive learning framework adapts a speech-trained HuBERT model to unlabeled violin recordings, achieving the best reported correlations with ...

A proof-of-concept study fused EEG recordings with clinical interview transcripts from 520 participants, achieving near-ceiling accuracy under controlled conditions while ...

A new open-source tool called THEMISim lets researchers search the THEMIS project's archive of more than a billion auroral images ...

Prototype-reliability calibration improved plant leaf recognition across three benchmarks and a laboratory-to-field test while leaving the exported classifier architecture unchanged.

HGACL-DRP achieved reported mean AUC values of 95.48% and 98.99% on two public cancer cell-line datasets in a computational drug-response ...