Deriving actionable insights from real-world data
The research in DataMine laboratory focuses on the development of novel tools and approaches for data-centric artificial intelligence (DCAI). Unlike model-centric artificial intelligence (MCAI), which treats training data as auxiliary to the learning process and spends extraordinary time on optimizing the parameters of the model, DCAI aims to achieve better outcomes by keeping the model and its parameters unchanged and spends more time on improving the quality of the training data. Using non-trivial and automated data selection approaches, in our past research, we showed that accurate models may be trained with significantly less data, thus, requiring fewer computing resources and reducing the carbon footprint of MCAI.
We are currently recruiting motivated and creative undergraduate and graduate students for two projects:
link to personal page
Michael is developing novel approaches for core-set selection for machine learning.
Shelton is working on evoluationary multi-objective optimization.
Get in touch, if you are interested in collaborations, have project ideas, or want to discuss our research.