An Intuitive Visual Interpretability For Convolutional Neural Networks

The first convolutional neural network was the Time Delay Neural Network (TDNN) proposed by Alexander Waibel in 1987 [5]. TDNN is a convolutional neural network applied to speech recognition problems. It uses FFT preprocessed speech signals as input. Its hidden layer consists of two one-dimensional convolution kernels to extract translation-invariant features in the frequency domain [6]. Before the advent of TDNN, the field of artificial intelligence made breakthrough progress in the research of back-propagation (BP) [7], so TDNN was able to use the BP framework for learning.