Tag: time-series forecasting

A physics-guided graph neural network with an extended LSTM backbone achieved year-long daily runoff forecasts on the middle Yellow River, ...

Researchers trained a deep learning model to forecast water scarcity characterisation factors for nearly 10,000 river basins through 2032, revealing ...

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

A prospective pilot study at a Turkish children's hospital shows that pairing deep learning demand forecasts with optimization-based physician scheduling ...

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

Researchers at Shanxi University have developed MAST-LLM, a framework that repurposes large language models with mixture-of-experts alignment and bidirectional spatio-temporal ...

A new lightweight deep learning model called BOA-LSTM combines Bayesian optimization and attention mechanisms to deliver faster and more accurate ...

Researchers used seven machine-learning models, led by Transformer and LSTM architectures, to reconstruct deep soil temperatures in cold-region canal slopes ...

A hybrid NARX–XGBoost machine learning framework predicts daily milk yields in heat-stressed dairy cows with substantially higher accuracy than existing ...

Researchers have developed a graph neural network model that learns dynamic time delays between variables in multivariate time series, improving ...