{"ID":2841738,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2511.11781","arxiv_id":"2511.11781","title":"Coordinate Descent for Network Linearization","abstract":"ReLU activations are the main bottleneck in Private Inference that is based on ResNet networks. This is because they incur significant inference latency. Reducing ReLU count is a discrete optimization problem, and there are two common ways to approach it. Most current state-of-the-art methods are based on a smooth approximation that jointly optimizes network accuracy and ReLU budget at once. However, the last hard thresholding step of the optimization usually introduces a large performance loss. We take an alternative approach that works directly in the discrete domain by leveraging Coordinate Descent as our optimization framework. In contrast to previous methods, this yields a sparse solution by design. We demonstrate, through extensive experiments, that our method is State of the Art on common benchmarks.","short_abstract":"ReLU activations are the main bottleneck in Private Inference that is based on ResNet networks. This is because they incur significant inference latency. Reducing ReLU count is a discrete optimization problem, and there are two common ways to approach it. Most current state-of-the-art methods are based on a smooth appr...","url_abs":"https://arxiv.org/abs/2511.11781","url_pdf":"https://arxiv.org/pdf/2511.11781v1","authors":"[\"Vlad Rakhlin\",\"Amir Jevnisek\",\"Shai Avidan\"]","published":"2025-11-14T14:03:58Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.CV\",\"stat.ML\"]","methods":"[]","has_code":false}