mlsanityData to 13 Oct 2026

Selective Network Linearization for Efficient Private Inference

arXiv 2202.02340 ↗First posted 4 Feb 2022cs.CR cs.LGcited by 31

Not on the map. Why?

Abstract (arXiv)

Private inference (PI) enables inference directly on cryptographically secure data.While promising to address many privacy issues, it has seen limited use due to extreme runtimes. Unlike plaintext inference, where latency is dominated by FLOPs, in PI non-linear functions (namely ReLU) are the bottleneck. Thus, practical PI demands novel ReLU-aware optimizations. To reduce PI latency we propose a gradient-based algorithm that selectively linearizes ReLUs while maintaining prediction accuracy. We evaluate our algorithm on several standard PI benchmarks. The results demonstrate up to 4.25% more accuracy (iso-ReLU count at 50K) or 2.2× less latency (iso-accuracy at 70%) than the current state of the art and advance the Pareto frontier across the latency-accuracy space. To complement empirical results, we present a "no free lunch" theorem that sheds light on how and when network linearization is possible while maintaining prediction accuracy. Public code is available at urlhttps://github.com/NYU-DICE-Lab/selectivenetworklinearization.

Key facts. Cited by 31 papers in the map (data through 13 Oct 2026). Builds on 17 of its 30 references that are in the map. First posted on arXiv 4 Feb 2022.

Builds on

17 of 30 references.

References · 17

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,950
  2. 2015Distilling the Knowledge in a Neural Networkcited by 10,180
  3. 2015Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classificationcited by 3,601
  4. 2016Wide Residual Networkscited by 3,063
  5. 2016Pruning Filters for Efficient ConvNetscited by 1,138
  6. 2018SNIP: Single-shot Network Pruning based on Connection Sensitivitycited by 809
  7. 2017To prune, or not to prune: exploring the efficacy of pruning for model compressioncited by 602
  8. 2019The State of Sparsity in Deep Neural Networkscited by 474
  9. 2018Filter Pruning via Geometric Median for Deep Convolutional Neural Networks Accelerationcited by 367
  10. 2015Representation Benefits of Deep Feedforward Networkscited by 85
  11. 2020CrypTFlow2: Practical 2-Party Secure Inferencecited by 82
  12. 2018Small ReLU networks are powerful memorizers: a tight analysis of memorization capacitycited by 77
5 more
  1. 2021DeepReDuce: ReLU Reduction for Fast Private Inferencecited by 50
  2. 2020CryptoNAS: Private Inference on a ReLU Budgetcited by 40
  3. 2021Circa: Stochastic ReLUs for Private Deep Learningcited by 16
  4. 2021Sphynx: ReLU-Efficient Network Design for Private Inferencecited by 15
  5. 2020ESPN: Extremely Sparse Pruned Networkscited by 3

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2022: 1; 2023: 10; 2024: 13; 2025: 4; 2026: 3.

31 papers.

Sort:

By area:

  1. 2023AutoReP: Automatic ReLU Replacement for Fast Private Network Inferencecited by 29
  2. 2023LinGCN: Structural Linearized Graph Convolutional Network for Homomorphically Encrypted Inferencecited by 27
  3. 2023Learning to Linearize Deep Neural Networks for Secure and Efficient Private Inferencecited by 20
  4. 2022MPCViT: Searching for Accurate and Efficient MPC-Friendly Vision Transformer with Heterogeneous Attentioncited by 19
  5. 2025Are We Truly Forgetting? A Critical Re-examination of Machine Unlearning Evaluation Protocolscited by 18
  6. 2024AdaPI: Facilitating DNN Model Adaptivity for Efficient Private Inference in Edge Computingcited by 15
  7. 2023RRNet: Towards ReLU-Reduced Neural Network for Two-party Computation Based Private Inferencecited by 12
  8. 2024PrivCirNet: Efficient Private Inference via Block Circulant Transformationcited by 7
  9. 2024MPC-Minimized Secure LLM Inferencecited by 5
  10. 2023Making Models Shallow Again: Jointly Learning to Reduce Non-Linearity and Depth for Latency-Efficient Private Inferencecited by 4
  11. 2023PriViT: Vision Transformers for Fast Private Inferencecited by 4
  12. 2024The Simpler The Better: An Entropy-Based Importance Metric To Reduce Neural Networks' Depthcited by 3
19 more
  1. 2023Approximating ReLU on a Reduced Ring for Efficient MPC-based Private Inferencecited by 2
  2. 2024HEQuant: Marrying Homomorphic Encryption and Quantization for Communication-Efficient Private Inferencecited by 2
  3. 2024Disparate Impact on Group Accuracy of Linearization for Private Inferencecited by 2
  4. 2023Securing Neural Networks with Knapsack Optimizationcited by 1
  5. 2023Towards Fast and Scalable Private Inferencecited by 1
  6. 2023LayerCollapse: Adaptive compression of neural networkscited by 1
  7. 2024Neural Networks with (Low-Precision) Polynomial Approximations: New Insights and Techniques for Accuracy Improvementcited by 1
  8. 2024PrivQuant: Communication-Efficient Private Inference with Quantized Network/Protocol Co-Optimizationcited by 1
  9. 2024TruncFormer: Private LLM Inference Using Only Truncationscited by 1
  10. 2025CrypTorch: PyTorch-based Auto-tuning Compiler for Machine Learning with Multi-party Computationcited by 1
  11. 2026On the (In-)Security of the Shuffling Defense in the Transformer Secure Inferencecited by 1
  12. 2024Linearizing Models for Efficient yet Robust Private Inferencecited by 0
  13. 2024xMLP: Revolutionizing Private Inference with Exclusive Square Activationcited by 0
  14. 2024Secure Multiparty Generative AIcited by 0
  15. 2024Impact of Privacy Parameters on Deep Learning Models for Image Classificationcited by 0
  16. 2025Coordinate Descent for Network Linearizationcited by 0
  17. 2025DeepShare: Sharing ReLU Across Channels and Layers for Efficient Private Inferencecited by 0
  18. 2026LRD-MPC: Efficient MPC Inference through Low-rank Decompositioncited by 0
  19. 2026UFO: Unlocking Ultra-Efficient Quantized Private Inference with Protocol and Algorithm Co-Optimizationcited by 0