Deep learning based anomaly and drift detection in Time series

In the ever-evolving landscape of machine learning, detecting concept drift and anomalies is crucial for maintaining model reliability and decision-making accuracy, particularly in dynamic real-world environments. Concept drift, where data distributions change over time, can significantly degrade model performance, while anomaly detection plays a vital role in identifying rare and potentially critical events across various domains. In this talk, I will present ADDM (Autoregressive-Based Drift Detection Method), a novel approach that effectively identifies and adapts to data shifts, ensuring models remain robust over time. Additionally, I will introduce AnoRand, a deep learning-based semi-supervised anomaly detection method that leverages synthetic label generation to enhance detection performance, even with limited labeled data. Our research demonstrates that both methods outperform state-of-the-art techniques across multiple datasets.
Beyond theoretical advancements, I will explore their practical applications in predictive maintenance, where these techniques enable early fault detection, optimized maintenance strategies, and reduced operational risks. Furthermore, I will discuss potential use cases in agriculture and marine biology, showcasing how AI-driven drift and anomaly detection can help monitor crop health, detect environmental changes, track marine species, and support conservation efforts. These applications highlight the transformative power of AI in ensuring sustainability and efficiency across diverse fields.