Gazelle: A Low Latency Framework for Secure Neural Network Inference · HackerPapers discussions

Gazelle: A Low Latency Framework for Secure Neural Network Inference — mentioned by advael

Mon, 10 Jun 2024 22:33:20 GMT

I'm disappointed that they make the fundamental claim that their cloud service is private with respect to user inputs passed through it and don't even a little bit talk about how that's accomplished. Even just an explanation of what guarantees they make and how would be much more interesting than explanations of their flavor of RLHF or whatever nonsense. I read the GAZELLE* paper when it came out and wondered what it would look like if a large-scale organization tried to deploy something like it. Of course, Apple will never give adequate details about security mechanisms or privacy…

Paper and discussion history. A mention is not an endorsement.

Gazelle: A Low Latency Framework for Secure Neural Network Inference — mentioned by testtest7777

Wed, 11 Jul 2018 19:51:00 GMT

The authors should do a comparison with recent work on HE for inference on other model types. Logistic regression is a subset of neural network inference, and there is significant prior work that appears to be competitive with their results. Recent work [0] shows that MNIST, for example, can be done in a model secure/data secure way with online 30ms inference latency. For ~500,000 samples, evaluated sequentially, that's roughly 5 hours, which beats their time by about 3x. Perhaps I'm missing details in my read of the paper. [0] Gazelle: A Low Latency Framework for Secure Neural Network…

Paper and discussion history. A mention is not an endorsement.