Topic
forecasting
Technology DeepMind's WeatherNext AI Predicts Hurricanes With an Extra Day of Lead Time
Google DeepMind and Google Research's WeatherNext model predicted Hurricane Melissa's Category 5 landfall in Jamaica with 80% confidence five days ahead. On average, it provides a day more lead time than existing forecast models, according to a paper in Nature. The extra day is critical for evacuations, staging supplies, and moving resources.
TelcoAgent: Foundation Model-Based Framework for Scalable and Explainable 5G KPM Forecasting
TelcoAgent is a foundation model-based framework for 5G key performance measurement forecasting that addresses scalability and explainability issues. It uses a 3GPP knowledge graph and time-series foundation model to deliver zero-shot predictions across multiple cells. Evaluated on a real-world city-scale 5G dataset from a U.S. operator, it achieved high accuracy for seven KPMs per cell across 200 cells.
FlowState: New Time-Series Model Handles Any Sampling Rate Without Retraining
IBM Research has developed FlowState, a novel time-series foundation model (TSFM) that is sampling-rate-equivariant, meaning it can handle data sampled at different rates without retraining. The model uses a state space encoder and a functional basis decoder to achieve continuous-time modeling, and it outperforms larger models on the GIFT-Eval benchmark while being one of the smallest TSFMs.
Phys-JEPA Model Promises More Accurate Multivariate Time-Series Forecasting with Physics-Informed Latent States
Phys-JEPA is a new architecture that imposes physical consistency on latent states rather than only on outputs, improving multivariate time-series forecasting. On standard benchmarks, it reduces mean squared error across multiple horizons, suggesting a promising direction for interpretable temporal world models.
RAID: Semantic Graph Diffusion Enables True Cold-Start and Cross-Lingual Forecasting
A new framework called RAID (Retrieval-Augmented Iterative Diffusion) addresses the true cold-start forecasting problem where no prior observations exist. By leveraging textual metadata and semantic graph diffusion, RAID outperforms strong foundation models on accuracy and prediction interval coverage while reducing inference latency by an order of magnitude. It also enables zero-shot cross-lingual transfer, allowing models trained in one language to generalize to others.
Green AI Carbon Optimizer Recommends Carbon-Efficient Training Locations and Forecasts Global AI Energy Demand
The Green AI Carbon Optimizer, presented in a new arXiv paper, offers two tools: a carbon-aware cloud region recommender for AI training and a power-law forecasting pipeline for global AI energy demand. By combining grid carbon intensity, renewable share, and PUE across 100+ regions, optimal region selection can reduce emissions by 97.2% versus the worst region. The forecasting model, based on 26 anchor models, projects 2030 AI energy demand between 7 TWh and 1,436 TWh depending on scenario assumptions.
TimeVista: Researchers Use Vision-Language Models as Judges for Time Series Forecasting Evaluation
Researchers propose using vision-language models (VLMs) as judges for time series forecasting, addressing limitations of traditional point-wise metrics. They introduce TimeVista, a benchmark of 5,563 samples, and show VLMs achieve significantly higher consistency with human preferences than conventional metrics, also assessing Time Series Foundation Models.
New AI Framework SERAF Combines Semantic and Numerical Data for Better Time Series Forecasting
Researchers propose SERAF, a semantics-enhanced retrieval-augmented time series forecasting framework that combines numerical similarity with textual descriptions to improve predictions under non-stationarity. The approach outperforms state-of-the-art baselines across seven real-world datasets.