About this Abstract |
| Meeting |
MS&T25: Materials Science & Technology
|
| Symposium
|
Enhancing the Accessibility of Machine Learning-Enabled Experiments
|
| Presentation Title |
Foundational Workflows for Processing Legacy Data and Realizing Domain-Specific Multi-Modal AI Models |
| Author(s) |
Kevin Le, Andrew Walter Richards, Robert Errol Hackenberg |
| On-Site Speaker (Planned) |
Kevin Le |
| Abstract Scope |
Efforts to transfer expert knowledge to the next generation is fraught with inefficiency due to several factors including: inability to capture all facets of knowledge, limited access to predecessor's time, and hindrances posed by their time and ability to parse and understand large amounts of unorganized knowledge contained in heritage files. Although legacy reports are the most promising with a view toward knowledge capture, they nonetheless suffer from limitations such as being born analog, lacking keywords, and having limited metadata context. There is a need to extract knowledge efficiently from these reports to aid human learning as well as to feed emerging AI tools. Therefore, researchers aim to develop tools that make legacy data useful for current needs and to prepare it for incorporation into domain-specific AI models. This talk discusses workflows for automated meta data generation to improve accessibility, organization, and understanding. |