The Research archive provides access to all Research articles published in past issues of Communications of the ACM.
In this work, we initiate a principled study of how to guarantee the validity of statistical inference in adaptive data analysis.
"Certifying a File System Using Crash Hoare Logic: Correctness in the Presence of Crashes" presents a big step toward real-world file systems that are crash-safe in a strict sense.
This paper introduces Crash Hoare logic, which extends traditional Hoare logic with a crash condition, a recovery procedure, and logical address spaces for specifying disk states at different abstraction levels.
The authors of "Guilt-Free Data Reuse" show there is a way to construct a safety net that goes around a fixed dataset so that it may be analyzed interactively without compromising statistical validity even when the dataset is…