Tag: transformer models

Researchers converted patent abstracts into semantic embeddings to measure technological similarity between firms and found the resulting network predicts subsequent ...

A new multi-domain Arabic dataset and benchmark shows that deepfake-text detectors excel within familiar domains but falter sharply on government ...

Researchers have unveiled a collaborative encryption framework that trains deep learning models on sensitive data up to three times faster ...

Researchers in India have built a custom fine-tuned BERT-based framework that outperforms classical machine learning, deep learning baselines, and GPT-3.5 ...

A systematic benchmark of classical, transformer-based and foundation AI models finds that a hybrid CNN-foundation-transformer architecture more than doubles the ...

Researchers have built a hybrid LSTM-BiLSTM deep learning model that adapts to seasonal wind patterns and outperforms transformers while using ...

A harmonised benchmark of lexical models, biomedical transformers and retrieval-grounded large language models shows that internal test scores fail to ...

Researchers have enhanced the TimeXer forecasting model with a temporal convolutional network and differential attention, achieving lower stock prediction errors ...

Researchers have developed SWIFT-KD, a federated learning framework combining hierarchical sliding-window transformers with knowledge distillation that predicts building energy consumption ...