Considerations of Integrating Computing-In-Memory and Processing-In-Sensor into Convolutional Neural Network Accelerators for Low-Power Edge Devices | IEEE Conference Publication | IEEE Xplore