Google I/O returns May 18-20! Reserve space and build your schedule Register now


Creates aggregator with compression and adaptive zeroing and clipping.

Zeroes out extremely large values for robustness to data corruption on clients, clips to moderately high norm for robustness to outliers. After weighting in mean, the weighted values are uniformly quantized to reduce the size of the model update communicated from clients to the server. For details, see Suresh et al. (2017) The default configuration is chosen such that compression does not have adverse effect on trained model quality in typical tasks.

zeroing Whether to enable adaptive zeroing.
clipping Whether to enable adaptive clipping.

A tff.aggregators.WeightedAggregationFactory.