TensorFlow as a Feature Engineering DSLDownload PDF

22 Sept 2019 (modified: 05 May 2023)Submitted to Program Transformations @NeurIPS2019Readers: Everyone
Keywords: feature engineering, dsl, tensorflow, keras, target-rate encoding, one-hot encoding
TL;DR: Frameworks for implementing Deep Learning networks can also be used as frameworks for feature engineering.
Abstract: Deep Neural Networks have fulfilled the promise to reduce the need of feature engineering by leveraging deep networks and training data. An interesting side effect, maybe unexpected, is that frameworks for implementing Deep Learning networks can also be used as frameworks for feature engineering. We sketch here some ideas for a DSL implemented on top of TensorFlow, hoping to attract interest from both the deep learning computational graph and programming languages communities.
3 Replies

Loading