| Presentation | 2023-02-28 Test Point Selection Method Using Graph Neural Networks and Deep Reinforcement Learning Shaoqi Wei, Kohei Shiotani, Senling Wang, Hiroshi Kai, Yoshinobu Higami, Hiroshi Takahashi, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | It is well known that selecting the optimal test point to maximize the fault coverage is NP-hard. Conventional heuristic algorithm may have limitations for the ever-complexing large-scale logic circuits. In this study, we propose a method named Deep-TP-Explorer for selecting test point insertion points using a deep graph-convolutional neural networks (GCN). In the method, a gate-level logic circuit is first represented by an undirected graph. The structural features including the gate type, connection relations, controllability/observability of the circuit are embedded into each node using GCN. The combination of signal lines that may contribute to improving the random testability is then generated using a deep neural network called TP Solver. To train the proposed TP Solver in a large-scale circuit, we applied an efficient reinforcement learning algorithm called Advantage Actor-Critic (A2C), which simultaneously trains an Actor (TP solver) to determine TPs’ combination and a Critic (value predictor) to evaluate the effect of TPs insertion (action). The effectiveness of the proposed method was confirmed on the ISCAS89 and ITC99 benchmark circuits. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Design for Test / Test point insertion / Reinforcement Deep Learning / Graph Neural Networks |
| Paper # | DC2022-87 |
| Date of Issue | 2023-02-21 (DC) |
| Conference Information | |
| Committee | DC |
|---|---|
| Conference Date | 2023/2/28(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Kikai-Shinko-Kaikan Bldg |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Tatsuhiro Tsuchiya(Osaka Univ.) |
| Vice Chair | Toshinori Hosokawa(Nihon Univ.) |
| Secretary | Toshinori Hosokawa(Nihon Univ.) |
| Assistant | |
| Paper Information | |
| Registration To | Technical Committee on Dependable Computing |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Test Point Selection Method Using Graph Neural Networks and Deep Reinforcement Learning |
| Sub Title (in English) | |
| Keyword(1) | Design for Test |
| Keyword(2) | Test point insertion |
| Keyword(3) | Reinforcement Deep Learning |
| Keyword(4) | Graph Neural Networks |
| 1st Author's Name | Shaoqi Wei |
| 1st Author's Affiliation | Ehime University(Ehime Univ.) |
| 2nd Author's Name | Kohei Shiotani |
| 2nd Author's Affiliation | Ehime University(Ehime Univ.) |
| 3rd Author's Name | Senling Wang |
| 3rd Author's Affiliation | Ehime University(Ehime Univ.) |
| 4th Author's Name | Hiroshi Kai |
| 4th Author's Affiliation | Ehime University(Ehime Univ.) |
| 5th Author's Name | Yoshinobu Higami |
| 5th Author's Affiliation | Ehime University(Ehime Univ.) |
| 6th Author's Name | Hiroshi Takahashi |
| 6th Author's Affiliation | Ehime University(Ehime Univ.) |
| Date | 2023-02-28 |
| Paper # | DC2022-87 |
| Volume (vol) | vol.122 |
| Number (no) | DC-393 |
| Page | pp.pp.27-32(DC), |
| #Pages | 6 |
| Date of Issue | 2023-02-21 (DC) |