Tolerating Soft Errors in Deep Learning Accelerators with Reliable On-Chip Memory Designs | IEEE Conference Publication | IEEE Xplore