Identifying and Controlling Important Neurons in Neural Machine Translation

Anthony Bau, Yonatan Belinkov, Hassan Sajjad, Nadir Durrani, Fahim Dalvi, James Glass

Abstract: Neural machine translation (NMT) models learn representations containing substantial linguistic information. However, it is not clear if such information is fully distributed or if some of it can be attributed to individual neurons. We develop unsupervised methods for discovering important neurons in NMT models. Our methods rely on the intuition that different models learn similar properties, and do not require any costly external supervision. We show experimentally that translation quality depends on the discovered neurons, and find that many of them capture common linguistic phenomena. Finally, we show how to control NMT translations in predictable ways, by modifying activations of individual neurons.

Keywords: neural machine translation, individual neurons, unsupervised, analysis, correlation, translation control, distributivity, localization

TL;DR: Unsupervised methods for finding, analyzing, and controlling important neurons in NMT