Jump to Content

Mike Bernico

Mike Bernico is a machine learning researcher and author specializing in synthetic data generation and differential privacy. His current research at Google involves developing novel synthetic data techniques for large-scale models like Gemini and implementing privacy-preserving solutions for products such as Workspace, Gmail, and Cloud.

Mike's contributions to the field are highlighted by four patents in machine learning and his book, "Deep Learning Quick Reference." He has a strong connection to academia, holding a master's degree from the University of Illinois at Springfield, where he also taught Data Science for six years. His applied research experience includes previous roles as an AI Engineer at Fitbit and Google Cloud, and leading the Data Science team at State Farm. Beyond his research, Mike is a photographer, traveler, and certified scuba diver.

Research Areas

Authored Publications
results

Publications

Years

Research Areas

Teams

Sort By
  • Title
  • Title, descending
  • Year
  • Year, descending
    Deep Learning Quick Reference
    Mike Bernico
    Packt Publishing (2018)
    Preview abstract Deep learning has become an essential necessity to enter the world of artificial intelligence. With this book deep learning techniques will become more accessible, practical, and relevant to practicing data scientists. It moves deep learning from academia to the real world through practical examples. View details
    Investigating the Impact of Data Volume and Domain Similarity on Transfer Learning Applications
    Mike Bernico
    2017
    Preview abstract Transfer learning allows practitioners to recognize and apply knowledge learned in previous tasks (source task) to new tasks or new domains (target task), which share some commonality. The two important factors impacting the performance of transfer learning models are: (a) the size of the target dataset, and (b) the similarity in distribution between source and target domains. Thus far, there has been little investigation into just how important these factors are. In this paper, we investigate the impact of target dataset size and source/target domain similarity on model performance through a series of experiments. We find that more data is always beneficial, and model performance improves linearly with the log of data size, until we are out of data. As source/target domains differ, more data is required and fine tuning will render better performance than feature extraction. When source/target domains are similar and data size is small, fine tuning and feature extraction renders equivalent performance. Our hope is that by beginning this quantitative investigation on the effect of data volume and domain similarity in transfer learning we might inspire others to explore the significance of data in developing more accurate statistical models. View details
    Search on Google Scholar
    ×