Research map / Federated learning and privacy / Privacy attacks and secure learning
Slalom: Fast, Verifiable and Private Execution of Neural Networks in Trusted Hardware
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Outsource linear layers and preprocess Freivalds checks | Slalom on VGG16 · VGG16 | throughput: up to a 20.3× increase | Helped |
| Outsource linear layers and verify with Freivalds | Slalom on MobileNet · MobileNet | throughput: 3.6× | Helped |
| Remove intermediate activations between depthwise and pointwise convolutions (ablation) | MobileNet · MobileNet, verifiable inference | throughput: 6.4× versus 3.6× | Helped |
| Blind outsourced layer inputs with precomputed factors | Slalom on VGG16 · VGG16 | throughput: 13.0× | Helped |
| Add integrity checks to private inference (ablation) | Slalom on VGG16 · VGG16 | throughput: 13.0× to 10.7× | Hurt |
| Add integrity checks to private inference (ablation) | Slalom on MobileNet (fused) · MobileNet (fused) | throughput: 5.0× to 4.1× | Hurt |
| Quantize model inputs and weights (ablation) | VGG16 · ImageNet validation | top 1 accuracy: 71.0 -> 70.6 | Hurt |
| Quantize model inputs and weights (ablation) | MobileNet · ImageNet validation | top 1 accuracy: 70.7 -> 70.5 | Hurt |
| Quantize model inputs and weights (ablation) | ResNet 50 · ImageNet validation | top 1 accuracy: 76.9 -> 76.4 | Hurt |
| Run Slalom outside SGX | Slalom on MobileNet and VGG16 · standard CPU | throughput: 3.9× and 8.1× | Helped |
Closest papers by meaning
- GroupCover: A Secure, Efficient and Scalable Inference Framework for On-device Model Protection based on TEEs (ICML 2024 · similarity 0.86)
- Certified but Private: Scalable Zero-Knowledge Proofs for the Formal Verification of Neural Networks (NeurIPS 2026 · similarity 0.85)
- BiShield-TEE: On the (In-)Security of Unilateral Weight Obfuscation in On-Device TEE-Shielded LLM Partition (NeurIPS 2026 · similarity 0.85)
- United We Stand: Accelerating Privacy-Preserving Neural Inference by Conjunctive Optimization with Interleaved Nexus (AAAI 2024 · similarity 0.84)
- FairProof : Confidential and Certifiable Fairness for Neural Networks (ICML 2024 · similarity 0.84)
- Relational DNN Verification With Cross Executional Bound Refinement (ICML 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.