Architecture Extraction on DNN Architecture Candidate Pool
100AccuracyYu et al.
Evaluation results
| Method | Links | |
|---|---|---|
| Yu et al.Side-Channel=Electromagnetic Trace, Candidate Size=15, Aggregate Data=✔2023.04 | 100 | |
| Hong et al.Side-Channel=Cache Timing Trace, Candidate Size=13, Aggregate Data=✗2023.04 | 100 | |
| Kumar Jha et al.Side-Channel=Memory, Timing, Power, and GPU Kernels, Candidate Size=15, Aggregate Data=✗2023.04 | 100 | |
| Liu et al.Side-Channel=Adversarial GPU Kernel Execution Latency, Candidate Size=5, Aggregate Data=✗2023.04 | 100 | |
| InferNetSide-Channel=GPU Kernel Profile, Candidate Size=90, Aggregate Data=✔2023.04 | 100 | |
| Patwari et al.Side-Channel=Memory, CPU, and GPU Usage Trace, Candidate Size=20, Aggregate Data=✗2023.04 | 99 | |
| Xiang et al.Side-Channel=Power Trace, Candidate Size=6, Aggregate Data=✔2023.04 | 96.5 |