The goal of any AI system is to augment and help human beings in meaningful ways and tasks. At OLIVES, we look beyond the traditional machine learning settings and allow learning to be more robust and efficient in terms of compute and data, more versatile in terms of being able to handle multiple problems and data modalities in realistic non-stationary scenarios. This is possible by adding constraints to the data collection and curation, and model selection, training, and validation processes that are inspired by real-world settings.
1.Physics-driven Learning in Seismic Applications
a.Semi-supervised Sequence Modeling for Elastic Impedance Inversion
2.Biomedical-driven Learning in Ophthalmic Applications
a.OLIVES Dataset: Ophthalmic Labels for Investigating Visual Eye Semantics
b.Biomedical applications of Multi-Modal Learning and Active Learning