Towards Low-Bit Quantization of Deep Neural Networks with Limited Data | IEEE Conference Publication | IEEE Xplore