Advanced Machine Learning in Time Series and Financial Forecasting

The recent developments in the research area of time series forecasting and financial market analysis have shown a significant shift towards leveraging advanced machine learning techniques, particularly in the context of multimodal data integration and causal discovery. Researchers are increasingly focusing on the fusion of textual, numerical, and temporal data to enhance predictive accuracy in both financial markets and environmental monitoring. The use of large language models (LLMs) and transformer architectures has become prominent, with innovations such as Prion-ViT and CausalStock demonstrating the potential of these models in capturing complex dependencies and causal relationships. Additionally, the field is witnessing a renaissance in architectural diversification, with hybrid models and diffusion models emerging as strong contenders. The integration of retrieval-augmented generation (RAG) techniques is also gaining traction, suggesting that dynamic and event-driven data can benefit significantly from context-aware forecasting methods. These advancements not only improve the precision of predictions but also offer new avenues for explainability and real-time decision-making in high-stakes environments such as financial trading and environmental safety. Notably, the development of fully automated forecasting frameworks and the exploration of dual-valued functions in causal emergence are pushing the boundaries of what is possible in these domains, promising more efficient and robust solutions for complex time series problems.

Sources

Influencers' Reposts and Viral Diffusion: Prestige Bias in Online Communities

Exploring Relationships Between Cryptocurrency News Outlets and Influencers' Twitter Activity and Market Prices

Machine learning-driven Anomaly Detection and Forecasting for Euclid Space Telescope Operations

A Comprehensive Survey of Time Series Forecasting: Architectural Diversity and Open Challenges

Fully Automated Correlated Time Series Forecasting in Minutes

Prion-ViT: Prions-Inspired Vision Transformers for Temperature prediction with Specklegrams

$\spadesuit$ SPADE $\spadesuit$ Split Peak Attention DEcomposition

DFT: A Dual-branch Framework of Fluctuation and Trend for Stock Price Prediction

Local vs. Global Models for Hierarchical Forecasting

Early Prediction of Natural Gas Pipeline Leaks Using the MKTCN Model

CausalStock: Deep End-to-end Causal Discovery for News-driven Stock Movement Prediction

MEANT: Multimodal Encoder for Antecedent Information

Multi-Modal Forecaster: Jointly Predicting Time Series and Textual Data

Ozone level forecasting in Mexico City with temporal features and interactions

EUR/USD Exchange Rate Forecasting incorporating Text Mining Based on Pre-trained Language Models and Deep Learning Methods

CryptoLLM: Unleashing the Power of Prompted LLMs for SmartQnA and Classification of Crypto Posts

Retrieval Augmented Time Series Forecasting

Dual-Valued Functions of Dual Matrices with Applications in Causal Emergence

Analyst Reports and Stock Performance: Evidence from the Chinese Market

Built with on top of