SMART uses artificial neural networks for travel behavior research


Transportation is an important part of daily life. With the surging population, the need to manage transportation effectively gains all the more significance. Seems like artificial intelligence can be added to the blend. The result of the efforts of a group of researchers at the Future Urban Mobility research group of Singapore MIT Alliance for Research and Technology(SMART), has brought forth the theory-based residual neural network(TB-ResNet). It is a synthetic framework that blends together discrete choice models(DCMs) and deep neural networks(DNNs). This will help in enhancing the decision-making analysis of individuals which plays a significant role in travel behavior research.


The details and information concerning TB-ResNet is clearly mentioned in the paper, “Theory-based residual neural networks: A synergy of discrete choice models and deep neural networks.” The paper was published recently in the journal, Transportation Research: Part B. 

Machine learning is being increasingly used in the transportation field. However, the two distinct concepts, DCMs and DNNs were usually viewed as opposing concepts.

What the TB-ResNet does is taking the best characteristics of both, and by synergizing them, a new strength is discovered to enrich the research findings. The simplicity of DCM and the expressive power of DNN is blended together to facilitate the generation of more accurate findings. The enhanced accuracy of the predictions thus generated will be of great help in conducting effective travel behavior research. TB-ResNet attests to the famous quote “Together we stand, divided we fall.” In comparison to DCM or DNN taken in isolation, the TB-ResNet displays improved predictiveness, interpretability, and robustness.

These characteristics gain all the more importance as these play a significant role in the smooth running of mobility companies, governments,a nd policy makers, helping them to facilitate the optimization of transport networks and to address the various challenges in the field of transportation. This makes transport planning more holistic and unified.

According to Associate Professor, Jinhua Zhao (SMART FM lead principal investigator),

“Our Future Urban Mobility research team focusses on developing new paradigms and innovating future urban mobility systems in and beyond Singapore. This new TB-ResNet framework is an important milestone that could enrich our investigations for impacts of decision-making models for urban development.”

The research is backed by the National Research Foundation(NRF) Singapore.