time-series-forecasting

ConceptTS introduces an interpretable forecasting framework that uses large language models to propose human-readable concepts for multivariate time-series prediction, achieving competitive accuracy with transparency through concept bottlenecks.

Fuzzy-MoE is a novel fuzzy logic-based Mixture-of-Experts model that improves non-stationary multivariate time series forecasting accuracy and interpretability through interpretable expert routing rules.

CLaST introduces a context-aware contrastive VAE framework for probabilistic multivariate time series forecasting, demonstrating significant performance improvements over baseline methods on multiple benchmarks.

This paper benchmarks nine deep learning models for energy forecasting on smart meter data, revealing that accuracy saturates with longer historical input, declines with extended prediction horizons, and lightweight models offer cost-effective alternatives.

This paper proposes MoFE, a novel deep learning framework integrating Fourier Neural Operators within a Mixture-of-Experts architecture to address challenges in cryptocurrency price forecasting, achieving state-of-the-art performance in Bitcoin price prediction.

ScenarioDiff is a hierarchical contextual reasoning framework for multimodal time series forecasting that organizes textual context into three levels to guide a Multimodal Diffusion Transformer, showing effectiveness in event-driven domains.

This paper compares six deep learning models for electricity price forecasting, establishing a standardized benchmark framework to enable consistent evaluation across markets, especially in low-data scenarios.

RouteTS is a unified forecasting framework that routes time series components between frequency and time domains based on spectral characteristics, improving accuracy and efficiency in handling periodicity and transience.

This paper introduces RAEF, a model-agnostic retrieval-augmented method for time series forecasting that improves accuracy and reduces computational overhead compared to fine-tuning approaches.

The paper identifies forecast collapse in time-series foundation models for hourly equity return prediction and introduces CalibRank to balance calibration and ranking, significantly improving cross-sectional correlation.

FM-LLM propose a frequency-enhanced mixture-of-experts framework that adapts frozen LLMs to time series forecasting without textual prompts, achieving state-of-the-art results by injecting spectral representations and separating periodic/non-periodic decoding.

This preprint compares XGBoost and LSTM for forecasting transmitted heat energy in District Heating Systems, finding that XGBoost consistently outperforms LSTM while offering lower computational cost and environmental impact.

REATS is a new approach that uses LLM reasoning to perform interpretable, sample-adaptive ensemble learning for time series forecasting. It combines textual and numerical features with chain-of-thought reasoning and a two-stage fine-tuning framework, outperforming competitive baselines on eight benchmarks.

PhysAttNet is a physics-informed attention framework that augments lightweight CNN forecasters with domain-guided regularization to improve accuracy and generalization in industrial and astrophysical time series forecasting.

KReF introduces a training-free retrieval framework for long-term time-series forecasting that constructs empirical predictive distributions from similar historical lookback-future pairs, achieving strong CRPS performance across multiple benchmarks.

This paper identifies an objective gap in Direct Forecasting for multivariate time series and proposes CvLoss, a plug-in structural regularizer that constrains forecast residuals on a cross-variable graph to improve consistency across synchronous and asynchronous interactions. Experiments show consistent improvements over competitive forecasting models.

CAMP introduces a cycle-aware multi-scale patch mixer for time series forecasting, achieving state-of-the-art results on multiple benchmarks through adaptive cycle learning and horizon-guided patch refinement.

CastFSR is a Fast–Slow–Reflect agentic reasoning framework that leverages LLMs for context-aware time series forecasting, combining fast lightweight forecasters, slow deliberative reasoning, and reflective evaluation to improve forecasting accuracy and consistency.

The paper presents NRFormer+, a spatio-temporal Transformer for forecasting nuclear radiation using atmospheric diffusion guidance, achieving state-of-the-art accuracy on two large-scale benchmarks from Japan.

This paper proposes DiffDiff, a diffusion framework for probabilistic time series forecasting that embeds predictability asymmetry into the diffusion trajectory, outperforming six diffusion baselines on seven benchmarks across four prediction horizons.