Image recognition and classification applications are often compute-bound. This negatively impacts deep learning inference throughput, thereby resulting in less than optimal user experiences.
For medical imaging, a customized Alexnet model was optimized on Intel® Optimization for Caffe*. Compared to FP32, Intel® Deep Learning Boost (Intel DL Boost) (delivered by Vector Neural Network Instructions (VNNI)/INT8) optimizations helped increase deep learning inference throughput by 2.02X (see chart)1 2 3, while meeting Neusoft’s accuracy requirements.
Significantly improved deep learning inference throughput, while meeting accuracy requirements, thus delivering better user experience.