Mitigating Gradient Inversion Risks in Language Models via Token Obfuscation.

Xinguo Feng, Zhongkui Ma, Zihan Wang, Alsharif Abuadbba, Guangdong Bai

Browse the full AsiaCCS paper archive.