Tag: transformer architecture

Researchers have developed a load-aware transformer architecture that achieves high detection accuracy and noise resilience for low-voltage alternating-current arc faults ...

A deep learning framework incorporating dynamic directed graph propagation achieved up to 89.21 percent diagnostic accuracy across large Alzheimer's and ...

Researchers adapted the lightweight DistilBERT language model using continued pre-training and retrieval techniques to significantly enhance transaction fraud detection over ...

A dual-module recommendation framework maps latent semantic correlations across knowledge graph relations and factorizes signed user-item interaction graphs to enhance ...

A comprehensive critical review synthesizes the mathematical principles, architectural innovations, and empirical scaling behaviors governing modern Transformer foundation models.

A new pose-guided transformer framework achieves 96.28% accuracy in recognizing tennis strokes from single frames by integrating cross-scale attention and ...

Researchers developed TDSPnet, a lightweight Transformer-based point cloud model that achieved 95.55 percent accuracy detecting welding defects in power battery ...

Researchers developed a bird's-eye-view architecture combining collaborative distillation and multi-domain attention to improve multimodal 3D object detection in complex driving ...

A tri-spectrogram fusion transformer called IAMPT achieves the highest accuracy, Macro-F1 and Macro-AUC among evaluated models for fine-grained multi-fault recognition ...

A transformer-based middle-fusion framework called T-3MS Fusion achieves 74.9% NDS and 72.8% mAP on nuScenes 3D object detection using a ...