Upcoming events

Switch to Calendar View

• Active learning for industrial applications: training machine learning models with less data

  • When
    4 Mar 2025
  •  
    10:00 AM

  Title:  Active learning for industrial applications: training machine learning models with less data

  Presenter: Davide Cacciarelli
  

  Time: March 4, 2025, 16:00-17:00 CET / 10:00 AM EST / 7:00 AM PST

  Abstract:  In this webinar, we will explore active learning-based sampling strategies and their potential applications to industrial data. Active learning is a valuable approach for streamlining the development of classification and regression models, particularly in environments where acquiring labelled data is both time-consuming and costly. By selecting the most informative data points to label, active learning enables high model performance with significantly reduced labelling efforts. We will begin with an overview of active learning, discussing key scenarios and query strategies. Following this, we will present a brief case study on the use of active learning to improve a defect detection classifier. The webinar will then shift focus to strategies for managing data streams, introducing stream-based active learning. We will cover techniques for training linear models, handling concept drift, and leveraging knowledge distillation for efficient model scaling. Join us to discover how active learning can enhance the development of machine learning models while minimising the need for expensive labelled data.

  Bio: Davide Cacciarelli is a Research Associate at Imperial College London’s Analytics & Markets Lab. His research lies at the intersection of industrial statistics and machine learning, focusing on improving the accuracy and reliability of monitoring and predictive models in industrial contexts. He holds a dual Ph.D. from the Technical University of Denmark and the Norwegian University of Science and Technology. His areas of expertise include active learning, design of experiments, and statistical process control. Currently, his work has focused on the intersection of these methodologies with electricity markets, leveraging causal inference to uncover the relationships between renewable energy integration and market dynamics.
  

  Register Here
  

  Show details