Kumawat, Sudhakar; Verma, Manisha; Nakashima, Yuta; Raman, Shanmuganathan
(Institute of Electrical and Electronics Engineers, 2021-04)
Conventional 3D convolutional neural networks (CNNs) are computationally expensive, memory intensive, prone to overfitting, and most importantly, there is a need to improve their feature learning capabilities. To address ...