Comments on the Proof of Adaptive Stochastic Set Cover Based on Adaptive Submodularity and Its Implications for the Group Identification Problem in “Group-Based Active Query Selection for Rapid Diagnosis in Time-Critical Situations”

We point out an issue with one of the results in Bellala et al.[1] that invokes a main result on adaptive stochastic minimum cost cover problem (Theorem 5.8) of Golovin and Krause. We construct an example that shows that the proof of Theorem 5.8 of Golovin and Krause is invalid, and therefore, the proof in Bellala et al. about the near-optimum performance of their algorithm is also invalid.