Speaker: Dr. Maryam Bagherian
From: Yale University
Abstract
Distance metric learning is an approach in high-dimensional analysis which facilitates the exploration of underlying structures among data points in an effective and efficient manner. By learning a suitable distance metric, distance-based algorithms can better capture the intrinsic structure of data points, leading to improved performance in various tasks.
Multi-metric learning and geometric metric learning, in contrast to single-metric learning approaches, have demonstrated higher efficiency in handling complex data distributions or diverse data characteristics. These approaches offer increased flexibility and interpretability, and are particularly helpful in representation learning for complex multi-modal datasets.
In this context, we provide a brief introduction to the concepts of metric learning mentioned above, and propose an algorithm under metric learning constraints. Furthermore, we highlight examples of where these approaches can be applied to enhance performance and interpretability in various machine learning tasks.
For more info, please follow this link.
Read More