Challenges and Future Directions for Accountable Machine Learning

In recent years, machine learning (ML) algorithms have been applied in many areas such as healthcare, finance and autonomous vehicles. At the same time, there is an increasing need for making ML systems accountable, which would help deal with situations when these systems made wrong decisions or predictions. Currently there exist three major frameworks for Accountable ML: Model Card Toolkit, Datasheets, and FactSheets. However, the greatest limitation of these frameworks is that they are mostly focusing on qualitative information about the machine learning models. In this research, we discuss in detail these three frameworks and future development directions of Accountable ML frameworks; we recommend the implementation of causality, decision provenance, and computational tests for achieving better ML accountability.