Machine Learning Supply Chain Security
Primary Investigator:
Jamie Davis
Taylor R. Schorlemmer, Wenxin Jiang, James C. Davis
Abstract
This poster attempts to summarize some of the key issues in the Machine Learning (ML) supply chain. First, the poster discusses the elements of a traditional software supply chain, demonstrates a software supply chain attack pattern, proposes principles for a secure supply chain, and applies those principles to current security techniques. Next, it shows how an ML supply chain is formed by the reliance on pre-trained models. Finally, it shows potential supply chain threats to model hubs and discusses future work to mitigate risks of attack.