Advances in Machine Learning and Adaptive Systems

The field of machine learning and adaptive systems is rapidly evolving, with a focus on developing innovative methods to address complex challenges such as concept drift, data heterogeneity, and security threats. Researchers are exploring new approaches to improve the accuracy and efficiency of machine learning models, including the use of pseudo-labeling, online learning, and ensemble methods. Additionally, there is a growing interest in developing adaptive systems that can learn from data streams and adapt to changing conditions in real-time. Notable papers in this area include:

  • ADAPT, which introduces a novel pseudo-labeling semi-supervised algorithm for addressing concept drift in malware detection.
  • IncA-DES, which proposes an incremental and adaptive dynamic ensemble selection approach for data streams with concept drift.
  • OL-MDISF, which constructs a latent copula-based representation for heterogeneous features and detects drifts via ensemble entropy and latent mismatch. These advancements have the potential to significantly improve the performance and robustness of machine learning models and adaptive systems, enabling them to better handle complex and dynamic real-world scenarios.

Sources

Maneuver Detection via a Confidence Dominance Maneuver Indicator

ADAPT: A Pseudo-labeling Approach to Combat Concept Drift in Malware Detection

Dually Hierarchical Drift Adaptation for Online Configuration Performance Learning

Extension OL-MDISF: Online Learning from Mix-Typed, Drifted, and Incomplete Streaming Features

Distributed Resilient State Estimation and Control with Strategically Implemented Security Measures

Learning, fast and slow: a two-fold algorithm for data-based model adaptation

IncA-DES: An incremental and adaptive dynamic ensemble selection approach using online K-d tree neighborhood search for data streams with concept drift

Learning-Based Cost-Aware Defense of Parallel Server Systems against Malicious Attacks

Built with on top of