CrypTen: Secure Multi-Party Computation Meets Machine Learning
Authors: Brian Knott, Shobha Venkataraman, Awni Hannun, Shubho Sengupta, Mark Ibrahim, Laurens van der Maaten
NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our benchmarks show that CRYPTEN s GPU support and high-performance communication between (an arbitrary number of) parties allows it to perform efficient private evaluation of modern machine-learning models under a semi-honest threat model. For example, two parties using CRYPTEN can securely predict phonemes in speech recordings using Wav2Letter [17] faster than real-time. We hope that CRYPTEN will spur adoption of secure MPC in the machine-learning community. We performed benchmark experiments on a proprietary cluster, testing inference on both CPUs (Intel Skylake 18-core 1.6GHz) and GPUs (n Vidia P100). |
| Researcher Affiliation | Industry | Brian Knott Shobha Venkataraman Awni Hannun Shubho Sengupta Mark Ibrahim Laurens van der Maaten Facebook AI Research {brianknott,shobha,awni,ssengupta,marksibrahim,lvdmaaten}@fb.com |
| Pseudocode | No | The paper includes Python code examples in figures but no structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Code reproducing the results of our experiments is available on https://crypten.ai. |
| Open Datasets | Yes | We performed text-sentiment classification experiments on the Yelp review dataset [70] using a model that consists of a linear layer operating on word embeddings. We performed speech-recognition experiments using Wav2Letter [17] on the Libri Speech dataset [53]. We performed image-classification experiments on the Image Net dataset using residual networks (Res Nets; [35]) and vision transformers (Vi T; [26]). |
| Dataset Splits | No | The paper mentions using well-known datasets but does not explicitly provide the training, validation, and test dataset splits (e.g., percentages or sample counts) used for their experiments. |
| Hardware Specification | Yes | We performed benchmark experiments on a proprietary cluster, testing inference on both CPUs (Intel Skylake 18-core 1.6GHz) and GPUs (n Vidia P100). |
| Software Dependencies | No | The paper mentions using Py Torch, CUDA libraries such as cu BLAS and cu DNN, but it does not specify concrete version numbers for these software dependencies. |
| Experiment Setup | Yes | We set the number of Open MP threads to 1 in all benchmarks. All experiments were performed with the parties running in separate processes on a single machine. For GPU experiments, each party was assigned its own GPU. We evaluated the model on GPUs, varying the batch size and the number of parties participating. We experimented with a Res Net-18 with 18 convolutional layers and with a Vi T-B/16 model. Following common practice [35], we preprocess images by rescaling them to size 256 256 and taking a center crop of size 224 224. |