Delen Twitter Facebook LinkedIn Enhancing AI Learning: Smarter, fairer, and more robust solutions

Machine learning has revolutionized the way AI models are trained, but traditional methods rely heavily on pre-labelled datasets, which can be costly, time-consuming, and sometimes prone to bias. Current approaches often fall short in real-world scenarios—whether it’s handling fairness, working with graph data, defending against adversarial attacks, or managing imbalanced datasets. Active learning offers a smarter alternative by focusing only on the most valuable data points, helping AI learn faster with fewer labeled samples.
 
However, implementing active learning in these complex environments presents its own set of difficulties. To address these challenges, PhD researcher Ricky Fajri developed innovative methods to make active learning safer, fairer, and more effective across these demanding contexts. He defended his thesis on Tuesday, February 4.

[....]