[1] Lee, Taegyeong, et al. "Occlumency: Privacy-preserving remote deep-learning inference using SGX." The 25th Annual International Conference on Mobile Computing and Networking. 2019.
[2] Götzfried, Johannes, et al. "Cache attacks on Intel SGX." Proceedings of the 10th European Workshop on Systems Security. 2017.
[3] Murdock, Kit, et al. "Plundervolt: Software-based fault injection attacks against Intel SGX." 2020 IEEE Symposium on Security and Privacy (SP). IEEE, 2020.
[4] Sun, Zhichuang, et al. "ShadowNet: A Secure and Efficient On-device Model Inference System for Convolutional Neural Networks." 2023 IEEE Symposium on Security and Privacy (SP). IEEE Computer Society, 2022.
[5] Shen, Tianxiang, et al. "{SOTER}: Guarding Black-box Inference for General Neural Networks at the Edge." 2022 USENIX Annual Technical Conference (USENIX ATC 22). 2022.
[6] Khandaker, Mustakimur Rahman, et al. "COIN attacks: On insecurity of enclave untrusted interfaces in SGX." Proceedings of the Twenty-Fifth International Conference on Architectural Support for Programming Languages and Operating Systems. 2020.
[7] Tramer, Florian, and Dan Boneh. "Slalom: Fast, verifiable and private execution of neural networks in trusted hardware." arXiv preprint arXiv:1806.03287 (2018).
[8] Asvadishirehjini, Aref, Murat Kantarcioglu, and Bradley Malin. "Goat: Gpu outsourcing of deep learning training with asynchronous probabilistic integrity verification inside trusted execution environment." arXiv preprint arXiv:2010.08855 (2020).
[9] Cai, Yi, et al. "Enabling Secure in-Memory Neural Network Computing by Sparse Fast Gradient Encryption." ICCAD. 2019.
[10] Lin, Ning, et al. "Chaotic weights: A novel approach to protect intellectual property of deep neural networks." IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems 40.7 (2020): 1327-1339.
[11] Juvekar, Chiraag, Vinod Vaikuntanathan, and Anantha Chandrakasan. "{GAZELLE}: A low latency framework for secure neural network inference." 27th USENIX Security Symposium (USENIX Security 18). 2018.
[12] Jie, Yixin, et al. "Multi-Party Secure Computation with Intel SGX for Graph Neural Networks." ICC 2022-IEEE International Conference on Communications. IEEE, 2022.
[13] Leyton-Brown, Kevin, and Yoav Shoham. "Essentials of game theory: A concise multidisciplinary introduction." Synthesis lectures on artificial intelligence and machine learning 2.1 (2008): 1-88.
| Work | Platform | Computation Overhead | Attacks on Edge Device | Model Confidentiality | Data Confidentiality | Integrity Check | Reduction in Accuracy | Model Modification | Utility of Edge Accelerators | Physical Attack Consideration | Black Box Privacy Attacks Possibility |
|---|---|---|---|---|---|---|---|---|---|---|---|
| CryptoNets [3] | MLaaS | high | Leakage of Architecture Information. | Negative | High | False | Small | Significant | None | None | Yes |
| AutoPrivacy [4] | MLaaS | high | Leakage of Architecture Information. | Negative | High | False | Small | Significant | None | None | Yes |
| MiniONN [5] | MLaaS | high | Leakage of Architecture Information. | Negative | High | False | Small | Significant | None | None | Yes |
| GAZELLE [6] | MLaaS | high | Leakage of Architecture Information. | Negative | High | False | Small | Significant | None | None | Yes |
| DELPHI [7] | MLaaS | high | Leakage of Architecture Information. | Negative | High | False | Small | Significant | None | None | Yes |
| SecDeep [8] | ARM Trustzone | medium | Network Completion[1,2] and Cold Boot Attack. Model Architecture is leaked through Side-channel attack. Side-channel attack [9], bus probing | Partial | High | True | None | None | Yes | No | Yes |
| GaurdiaNN [10] | Edge Devices with TEE and SRAM | medium | Model architecture leakage through bus probing (CPU-DRAM), SRAM compromise [13] | Full | High | False | None | None | Yes (Cryptographic) | Yes | Yes |
| DarkneTZ [11] | Edge Devices with TEE | low | Physical Attacks (bus probing), Network Completion [1,2]. TEE compromise | Partial | High | False | None | None | No | No | Partially Yes |
| PUF-PIM [12] | Edge Devices with SRAM | medium | SRAM compromise [13] | Full | None | False | None | none | Yes | No | Yes |
| Occlumency [14] | MLaaS (Cloud Intel SGX) | medium | SGX vulnerabilities (like side channel, roll back and DoS) | Partial | High | Yes | None | None | No | No | No |
| ShadowNet [15] | Edge Device with TEE | medium | SGX side channel attacks | Partial | No | No | None | Small | Yes | Yes | No |
| SOTER [16] | Edge Devices with TEE | medium | SGX Vulnerabilities | Partial | Yes | Yes | Small | Small | Yes | Yes | Yes |
| ---------- | --------------------- | ------ | -------------------------------------------------------------------- | ------- | --- | --- | ----- | ----- | --- | --- | --- |
| SFGE [17] | Edge devices NVM | low | Network Completion [1,2], Prone to Side Channel and Physical attacks | Full | No | No | None | Small | Yes | No | Yes |
[1] Tran, Cong, et al. "${\sf DeepNC} $ DeepNC: Deep Generative Network Completion." IEEE transactions on pattern analysis and machine intelligence 44.4 (2020): 1837-1852.
[2] Kim, Myunghwan, and Jure Leskovec. "The network completion problem: Inferring missing nodes and edges in networks." Proceedings of the 2011 SIAM international conference on data mining. Society for Industrial and Applied Mathematics, 2011.
[3] Gilad-Bachrach, Ran, et al. "Cryptonets: Applying neural networks to encrypted data with high throughput and accuracy." International conference on machine learning. PMLR, 2016.
[4] Lou, Qian, Song Bian, and Lei Jiang. "Autoprivacy: Automated layer-wise parameter selection for secure neural network inference." Advances in Neural Information Processing Systems 33 (2020): 8638-8647.
[5] Liu, Jian, et al. "Oblivious neural network predictions via minionn transformations." Proceedings of the 2017 ACM SIGSAC conference on computer and communications security. 2017.
[6] Juvekar, Chiraag, Vinod Vaikuntanathan, and Anantha Chandrakasan. "{GAZELLE}: A low latency framework for secure neural network inference." 27th {USENIX} Security Symposium ({USENIX} Security 18). 2018.
[7] Srinivasan, Wenting Zheng, P. M. R. L. Akshayaram, and Popa Raluca Ada. "DELPHI: A cryptographic inference service for neural networks." Proc. 29th USENIX Secur. Symp. 2019.
[8] Liu, Renju, et al. "Secdeep: Secure and performant on-device deep learning inference framework for mobile and iot devices." Proceedings of the International Conference on Internet-of-Things Design and Implementation. 2021.
[9] Hua, Weizhe, Zhiru Zhang, and G. Edward Suh. "Reverse engineering convolutional neural networks through side-channel information leaks." Proceedings of the 55th Annual Design Automation Conference. 2018.
[10] Choi, Jinwoo, et al. "GuardiaNN: Fast and Secure On-Device Inference in TrustZone Using Embedded SRAM and Cryptographic Hardware." Proceedings of the 23rd conference on 23rd ACM/IFIP International Middleware Conference. 2022.
[11] Mo, Fan, et al. "Darknetz: towards model privacy at the edge using trusted execution environments." Proceedings of the 18th International Conference on Mobile Systems, Applications, and Services. 2020.
[12] Li, Wen, et al. "Leveraging Memory PUFs and PIM-based encryption to secure edge deep learning systems." 2019 IEEE 37th VLSI Test Symposium (VTS). IEEE, 2019.
[13] Mahmod, Jubayer, and Matthew Hicks. "SRAM has no chill: exploiting power domain separation to steal on-chip secrets." Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems. 2022.
[14] Lee, Taegyeong, et al. "Occlumency: Privacy-preserving remote deep-learning inference using sgx." The 25th Annual International Conference on Mobile Computing and Networking. 2019.
[15] Sun, Zhichuang, et al. "ShadowNet: A Secure and Efficient On-device Model Inference System for Convolutional Neural Networks." 2023 IEEE Symposium on Security and Privacy (SP). IEEE Computer Society, 2022.
[16] Shen, Tianxiang, et al. "{SOTER}: Guarding Black-box Inference for General Neural Networks at the Edge." 2022 USENIX Annual Technical Conference (USENIX ATC 22). 2022.
[17] Cai, Yi, et al. "Enabling secure in-memory neural network computing by sparse fast gradient encryption." 2019 IEEE/ACM International Conference on Computer-Aided Design (ICCAD). IEEE, 2019.
[18] Salehi, Mohsen, and Karthik Pattabiraman. "Poster AutoPatch: Automatic Hotpatching of Real-Time Embedded Devices." Proceedings of the 2022 ACM SIGSAC Conference on Computer and Communications Security. 2022.
[1] Rakin , Adnan Siraj, et al. Deepsteal : Advanced model extractions leveraging efficient weight stealing in 2022 IEEE Symposium on Security and Privacy (SP) SP). IEEE, 2022
[2] Zhu, Yuankun , et al. "Hermes Attack: Steal DNN Models with Lossless Inference USENIX Security Symposium .
[3] Yan, Mengjia , Christopher Fletcher, and Josep Torrellas . "Cache telepathy: Leveraging shared resource attacks to learn DNN USENIX Security Symposium .
[4] Shan, Shawn, et al. "Post breach recovery: Protection against white box adversarial examples for leaked DNN models." Proceedings of the 2022 ACM SIGSAC Conference on Computer and Communications Security .
[5] Mo, Fan, et al. " Darknetz : towards model privacy at the edge using trusted execution Proceedings of the 18th International Conference on Mobile Systems, Applications, and Services .
[6] Hashemi, Hanieh , Yongqin Wang, and Murali Annavaram . DarKnight : An accelerated framework for privacy and integrity preserving deep learning using trusted MICRO 54: 54th Annual IEEE/ACM International Symposium on Microarchitecture .
[7] Hu, Xing, et al. " Deepsniffer : A dnn model extraction framework based on learning architectural hints." Proceedings of the Twenty Fifth International Conference on Architectural Support for Programming Languages and Operating Systems .
[8] Choi, Jinwoo , et al. GuardiaNN : Fast and Secure On Device Inference in TrustZone Using Embedded SRAM and Cryptographic Hardware." Proceedings of the 23rd conference on 23rd ACM/IFIP International Middleware Conference .
[9] Cao, Xiaoyu , Jinyuan Jia, and Neil Zhenqiang Gong. " IPGuard : Protecting intellectual property of deep neural networks via fingerprinting the classification Proceedings of the 2021 ACM Asia Conference on Computer and Communications Security .
[10] Hanzlik, Lucjan , et al. Mlcapsule : Guarded offline deployment of machine learning as a Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 2021.
[11] Kesarwani , Manish, et al. "Model extraction warning in mlaas paradigm." Proceedings of the 34th Annual Computer Security Applications Conference .
[12] Hou, Jiahui , et al. "Model Protection: Real time privacy preserving inference service for model privacy at the edge." IEEE Transactions on Dependable and Secure Computing 19.6 (2021): 4270 4284.
[13] Juuti , Mika, et al. "PRADA: protecting against DNN model stealing 2019 IEEE European Symposium on Security and Privacy ( EuroS&P )). IEEE,
[14] Zhang, Jialong , et al. "Protecting intellectual property of deep neural networks with Proceedings of the 2018 on Asia Conference on Computer and Communications Security .
[15] Sun, Zhichuang , et al. ShadowNet : A Secure and Efficient On device Model Inference System for Convolutional Neural Networks." 2023 IEEE Symposium on Security and Privacy (SP) SP). IEEE Computer Society, 2022.
[1] Lee, Taegyeong , et al. "Occlumency: Privacy preserving remote deep learning inference using sgx The 25th Annual International Conference on Mobile Computing and Networking .
[2] Kim, Kyungtae , et al. "Vessels: Efficient and scalable deep learning prediction on trusted Proceedings of the 11th ACM Symposium on Cloud Computing .
[3] Truong, Jean Baptiste, et al. "Memory efficient deep learning inference in trusted execution environments." 2021 IEEE International Conference on Cloud Engineering ( IC2E). IEEE,
[4] Shen, Tianxiang , et al. "{SOTER}: Guarding Black box Inference for General Neural Networks at the Edge." 2022 USENIX Annual Technical Conference (USENIX ATC 22) 22). 2022.
[5] Hua, Weizhe , et al. GuardNN : secure accelerator architecture for privacy preserving deep learning." Proceedings of the 59th ACM/IEEE Design Automation Conference . 2022.
[6] Li, Yuepeng , et al. "Lasagna: Accelerating secure deep learning inference in sgx enabled edge cloud." Proceedings of the ACM Symposium on Cloud Computing .
[7] Xiang, Yecheng , et al. Aegisdnn : Dependable and timely execution of dnn tasks with sgx 2021 IEEE Real Time Systems Symposium (RTSS) RTSS). IEEE,
[8] Guo, Yunqi , et al. "A model obfuscation approach to IoT 2021 IEEE Conference on Communications and Network Security (CNS) CNS). IEEE,
[9] Shrivastava, Nivedita, and Smruti R. Sarangi. " Seculator : A Fast and Secure Neural Processing arXiv preprint arXiv:2204.08951 (
[10] Hashemi, Hanieh , Yongqin Wang, and Murali Annavaram . DarKnight : An accelerated framework for privacy and integrity preserving deep learning using trusted hardware." MICRO 54: 54th Annual IEEE/ACM International Symposium on Microarchitecture .
[11] Niu , Yue, Ramy E. Ali, and Salman Avestimehr . "3LegRace: Privacy Preserving DNN Training over TEEs and GPUs." arXiv preprint arXiv:2110.01229
[12] Mo, Fan, et al. " Darknetz : towards model privacy at the edge using trusted execution Proceedings of the 18th International Conference on Mobile Systems, Applications, and Services .
[13] Tramer , Florian, and Dan Boneh . "Slalom: Fast, verifiable and private execution of neural networks in trusted arXiv preprint arXiv:1806.03287
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent