Active Learning

In data-driven modeling, there exist emphasis on big data which sometimes might not be readily available. Generally, researchers including me prefers to use transfer learning which leverages the existence of data from another domain to enhance modeling scenario with limited dataset. Instead of transfer learning, another methodolody called Active learning which attempt to achieve high accuracy using as few labeled instances as possible, thereby minimizing the cost of obtaining labeled data, could be used.

Author

Name	Title/Affilation	Email
Teslim Olayiwola	PhD candidate, Louisiana State University	[email protected]

Plan

Implement different AL Query strategies to possibly increase performance while using small data\resolve missing data issues on

• Protein adsorption on polymer brushes
• [] Material informatics for property predictions
• [] Electrochemical separation with large missing data.

References

Active Learning (img): https://www.linkedin.com/feed/update/urn:li:activity:7192594281209999360
Query strategies: https://github.com/ej0cl6/deep-active-learning/tree/master/query_strategies