Find a story
Search Spins
Search titles, summaries, and missing voices across published articles — press releases, announcements, and media coverage.
9 results for “time series”
GENADA: efficient generative time series adversarial attack framework
Researchers introduced GENADA, a new generative adversarial attack framework for time series models that reduces computational cost by generating perturbations in a single forward pass instead of iterative gradient updates.
Aug 14, 2026
CAMP: A Cycle-Aware Multi-Scale Patch Mixer for Time Series Forecasting
CAMP is a new time series forecasting model introduced on arXiv that adapts to variable cyclic patterns and multi-scale temporal dynamics per input window, outperforming prior methods on multiple long-term forecasting benchmarks.
Aug 6, 2026
Optimizing Transformer Neural Network for Real-Time Outlier Detection on FPGAs
Researchers propose an FPGA-optimized Transformer architecture for real-time financial time-series outlier detection, aiming to improve speed and stability of downstream data processing.
Jul 28, 2026
CARNet Cycle-Conditioned Core Aggregation and Redistribution for Multivariate Time Series Forecasting
CARNet is a new attention-free deep learning architecture for multivariate time series forecasting that incorporates global periodic structure into linear-complexity core-based modeling, showing consistent empirical improvements over transformer and non-attention baselines on public benchmarks.
Jul 27, 2026
LLM4EHR: Aligning Clinical Time Series with Medical Event Sequences via Large Language Models
Researchers introduced LLM4EHR, a new clinical foundation model that aligns electronic health record (EHR) event sequences with time-series physiological data using a domain-adapted large language model and transformer-based time-series encoder, aiming to improve generalizability and performance on ICU outcome prediction tasks.
Jul 20, 2026
Exogenous Dropout: A Simple, Strong Baseline for Corruption-Robust Time Series Forecasting with Covariates
A new training technique called 'exogenous dropout' improves robustness of time series forecasting models to corrupted or missing exogenous covariates without sacrificing clean-data accuracy, and is released as an open benchmark and baseline recommendation.
Jul 9, 2026
Evaluating Time Series Foundation Models for Electricity Price Forecasting: Contamination Risk, Distributional Shifts, and Covariate Dependence
Researchers introduce a two-dataset benchmarking framework to rigorously evaluate time series foundation models (TSFMs) for electricity price forecasting, revealing their competitive but context-dependent performance and identifying contamination risk and covariate dependence as critical evaluation challenges.
Jul 8, 2026
StateFlow: Dual-State Recurrent Modeling for Long-Horizon Time Series Forecasting
Researchers introduce StateFlow, a dual-state recurrent modeling framework for long-horizon time series forecasting.
Published Jul 2, 2026 · Analyzed Jul 5, 2026
EVOTS: Evolutionary Transformer Search for Time Series Forecasting
Researchers introduced EVOTS, an evolutionary neural architecture search framework that automatically discovers task-adaptive Transformer-like models for multivariate time-series forecasting, achieving competitive or improved MSE over fixed Transformer baselines on ETT benchmarks.
Published Jul 2, 2026 · Analyzed Jul 5, 2026