Publication | Closed Access
Trained Uniform Quantization for Accurate and Efficient Neural Network Inference on Fixed-Point Hardware
35
Citations
0
References
2019
Year
Unknown Venue
Threshold GradientsConvolutional Neural NetworkEngineeringMachine LearningHardware AlgorithmAutomatic QuantizationImage AnalysisData SciencePattern RecognitionSparse Neural NetworkQuantization ThresholdsEmbedded Machine LearningMachine VisionFixed-point HardwareComputer EngineeringComputer ScienceDeep LearningNeural Architecture SearchQuantization (Signal Processing)Model CompressionComputer VisionUniform Quantization
We propose a method of training quantization thresholds (TQT) for uniform symmetric quantizers using standard backpropagation and gradient descent. Contrary to prior work, we show that a careful analysis of the straight-through estimator for threshold gradients allows for a natural range-precision trade-off leading to better optima. Our quantizers are constrained to use power-of-2 scale-factors and per-tensor scaling of weights and activations to make it amenable for hardware implementations. We present analytical support for the general robustness of our methods and empirically validate them on various CNNs for ImageNet classification. We are able to achieve near-floating-point accuracy on traditionally difficult networks such as MobileNets with less than 5 epochs of quantized (8-bit) retraining. Finally, we present Graffitist, a framework that enables automatic quantization of TensorFlow graphs for TQT (available at this https URL ).