Automated Transfer Learning Model for Counterfeit IC Detection
Chaitanya Bhure, Geraldine Shirley Nicholas, Shajib Ghosh, Yadi Zhong, Fareena Saqib · 2022
The global electronic supply chain threat with counterfeit semiconductors has been rapidly increasing impacting a wide range of electronic systems. Therefore, detection and avoidance of counterfeit parts have become an important priority with several methods developed for evaluating the authenticity of the ICs. Developed machine learning-based models require extensive data points, computation, and timing resources for training the dataset. Thus, the proposed automated model detects counterfeit ICs from different image acquisition modalities using transfer learning technique with limited resources producing more accurate results. The proposed model includes different pre-trained models such as VGG16 and VGG19 vision models along with the Inception v3 model where a comparative analysis shows that VGG16 produces a robust and generalized prediction with 80% accuracy.