Residual Logit (ResLogit) models are a family of machine learning based fully interpretable choice models underdevelopment in the Laboratory of Innovations in Transportation (LiTrans). So far, two models are available for general consumption:
Wong, Melvin, and Bilal Farooq. "ResLogit: A residual neural network logit model for data-driven choice modelling." Transportation Research Part C: Emerging Technologies 126 (2021): 103050.
Kamal, Kimia, and Bilal Farooq. "Ordinal-ResLogit: Interpretable Deep Residual Neural Networks for Ordered Choices." Journal of Choice Modelling 50 (2024).
The original implementation of ResLogit was done using Theano in Python. However, as the development of Theano stopped, we have moved to PyTorch in Python as the active development platform. The old Theano version of the code can be found in arxiv.