Tweet from PennNETS
RT @Aaroth: Bagdasaryan and Shmatikov find that training using private SGD increases error disparities between over and under-represented g… June 7, 2019, 4:01 pm
RT @Aaroth: Bagdasaryan and Shmatikov find that training using private SGD increases error disparities between over and under-represented g… June 7, 2019, 4:01 pm
