Papers › ISyNet: Convolutional Neural Networks design for AI accelerator
ISyNet: Convolutional Neural Networks design for AI accelerator
Alexey Letunovskiy, Vladimir Korviakov, Vladimir Polovnikov, Anastasiia Kargapoltseva, Ivan Mazurenko, Yepan Xiong
In recent years Deep Learning reached significant results in many practical problems, such as computer vision, natural language processing, speech recognition and many others. For many years the main goal of the research was to improve the quality of models, even if the complexity was impractically high. However, for the production solutions, which often require real-time work, the latency of the model plays a very important role. Current state-of-the-art architectures are found with neural architecture search (NAS) taking model complexity into account. However, designing of the search space suitable for specific hardware is still a challenging task. To address this problem we propose a measure of hardware efficiency of neural architecture search space - matrix efficiency measure (MEM); a search space comprising of hardware-efficient operations; a latency-aware scaling method; and ISyNet - a set of architectures designed to be fast on the specialized neural processing unit (NPU) hardware and accurate at the same time. We show the advantage of the designed architectures for the NPU devices on ImageNet and the generalization ability for the downstream classification and detection tasks.
Code
Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.
Code Syntology ran Syntology
Not run by Syntology. Nothing on this page verifies that the listed code works.
Tasks
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Neural Architecture Search | ImageNet | ISyNet-N3 | Top-1 Error Rate | 19.84 | #20 of 135 | Archive leaderboard | report |
| Neural Architecture Search | ImageNet | ISyNet-N2 | Top-1 Error Rate | 20.59 | #32 of 135 | Archive leaderboard | report |
| Neural Architecture Search | ImageNet | ISyNet-N1-S3 | Top-1 Error Rate | 21.45 | #46 of 135 | Archive leaderboard | report |
| Neural Architecture Search | ImageNet | ISyNet-N1-S2 | Top-1 Error Rate | 22.15 | #57 of 135 | Archive leaderboard | report |
| Neural Architecture Search | ImageNet | ISyNet-N1-S1 | Top-1 Error Rate | 22.7 | #67 of 135 | Archive leaderboard | report |
| Neural Architecture Search | ImageNet | ISyNet-N1 | Top-1 Error Rate | 23.16 | #78 of 135 | Archive leaderboard | report |
| Neural Architecture Search | ImageNet | ISyNet-N0 | Top-1 Error Rate | 24.55 | #108 of 135 | Archive leaderboard | report |
Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections