Review for NeurIPS paper: A General Method for Robust Learning from Batches

Additional Feedback: From the writeup it is not clear how the techniques are related to prior work. For example: 1) The filtering framework of [J019], that the current work builds on, is itself an adaptation of the filtering technique developed in the robust statistics literature, [DKK 16], etc. It would be appropriate to explain this and also elaborate on the differences between [J019] and the current application of filtering (which is done to some extent on p. 8). The current paper builds on these ideas. It would be useful for the non-expert reader if this was clarified.