{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,27]],"date-time":"2026-05-27T18:27:22Z","timestamp":1779906442223,"version":"3.53.1"},"reference-count":0,"publisher":"EasyChair","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>Private inference on neural networks requires running all the computation on encrypted data. Unfortunately, neural networks contain a large number of non-arithmetic operations, such as ReLU activation functions and max pooling layers, which incur a high latency cost in their encrypted form. To address this issue, the majority of private inference methods re- place some or all of the non-arithmetic operations with a polynomial approximation. This step introduces approximation errors that can substantially alter the output of the neural network and decrease its predictive performance. In this paper, we propose a Lipschitz- Guided Abstraction Refinement method (LiGAR), which provides strong guarantees on the global approximation error. Our method is iterative, and leverages state-of-the-art Lipschitz constant estimation techniques to produce increasingly tighter bounds on the worst-case error. At each iteration, LiGAR designs the least expensive polynomial approx- imation by solving the dual of the corresponding optimization problem. Our preliminary experiments show that LiGAR can easily converge to the optimum on medium-sized neural networks.<\/jats:p>","DOI":"10.29007\/59w3","type":"proceedings-article","created":{"date-parts":[[2023,10,23]],"date-time":"2023-10-23T18:31:41Z","timestamp":1698085901000},"page":"35-22","source":"Crossref","is-referenced-by-count":3,"title":["Certified Private Inference on Neural Networks via Lipschitz-Guided Abstraction Refinement"],"prefix":"10.29007","volume":"16","author":[{"given":"Edoardo","family":"Manino","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bernardo","family":"Magri","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mustafa","family":"Mustafa","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lucas","family":"Cordeiro","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"11545","event":{"name":"Proceedings of the 6th Workshop on Formal Methods for ML-Enabled Autonomous Systems"},"container-title":["Kalpa Publications in Computing"],"original-title":[],"deposited":{"date-parts":[[2023,10,23]],"date-time":"2023-10-23T18:31:42Z","timestamp":1698085902000},"score":1,"resource":{"primary":{"URL":"https:\/\/easychair.org\/publications\/paper\/bVbP"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[null]]},"references-count":0,"URL":"https:\/\/doi.org\/10.29007\/59w3","relation":{},"ISSN":["2515-1762"],"issn-type":[{"value":"2515-1762","type":"print"}],"subject":[]}}