Do Convolutional Neural Networks act as Compositional Nearest Neighbors?
share
We present a simple approach based on pixel-wise nearest neighbors to understand and interpret the functioning of state-of-the-art neural networks for pixel-level tasks. We aim to understand and uncover the synthesis/prediction mechanisms of state-of-the-art convolutional neural networks. To this end, we primarily analyze the synthesis process of generative models and the prediction mechanism of discriminative models. The main hypothesis of this work is that convolutional neural networks for pixel-level tasks learn a fast compositional nearest neighbor synthesis/prediction function. Our experiments on semantic segmentation and image-to-image translation show qualitative and quantitative evidence supporting this hypothesis.
READ FULL TEXT
