Researchers in Future urban mobility (FM) in Interdisciplinary Research Group Singapore-MIT Alliance for Research and Technology (SMART), MIT’s research venture in Singapore, has created a synthetic framework known as a theory-based residual neural network (TB-ResNet), which allows discrete choice models (DCMs) and deep neural networks (DNNs). Combines, also known as intensive learning. , To improve the analysis of individual decision making used in travel behavior research.
In their paper, “Theory-based residual neural networks: discrete choice models and deep neural network synergy, “Recently published in the journal Transportation Research: Part B, Smart researchers explained their developed TB-ResNet framework and demonstrated the strength of the combination of DCM and DNN methods, proving that they are highly complementary.
As machine learning is increasingly used in the field of transportation, two dissimilar research concepts, DCM and DNN, have long been seen as conflicting methods of research.
By coordinating these two important research paradigms, TB-ResNet leverages the simplicity of DCM and the expressive power of DNN to make more accurate predictions for richer decision making and personal decision-making analysis, for better travel behavior research is important. The developed TB-ResNet framework is more predictive, interpretable, and robust than DCM or DNN, with findings consistent with several types of datasets.
Accurate and efficient analysis of individual decision-making in everyday contexts is important for mobility companies, governments and those seeking to optimize transport networks and especially to meet transportation challenges in cities. TB-Resnet will eliminate the current difficulties present in DCM and DNN and allow stakeholders to take a holistic, integrated approach to transportation planning.
Urban Mobility Lab at MIT’s Postdoc and lead author Shenhao Wang says, “Better insights for travelers to make decisions about travel modes, destinations, departure times and planning activities, urban transportation for governments and transportation companies around the world Are critical to the plan. I look forward to developing TB-Resnet and its applications for transportation planning, which is now accepted by the transportation research community. “
“Our Future Urban Mobility research team focuses on developing new paradigms within and outside Singapore and innovating future urban mobility systems,” says the MIT Department of Smart FM, Chief Principal Investigator and Associate Professor of Urban Studies and Planning, Jihua Zhao. is. This new TB-Resnet framework is an important milestone that may enrich our investigation for the implications of decision-making models for urban development. “
TB-Resnet can also be widely applied to understanding individual decision-making cases, as demonstrated in this research, whether it is about travel, consumption, or voting, among many others. between.
This study tested the TB-Resnet framework in three examples. First, researchers used it to predict travel mode decisions between transit, driving, autonomous vehicles, walking, and cycling, which are the major travel modes in an urban setting. Second, they evaluated risk options and preferences when uncertainty involves monetary repayment. Examples of such situations include insurance, financial investment, and voting decisions.
Finally, he examined the temporary option, which measures the tradeoff between current and future money payments. A typical example of such decisions being made would be in transportation development, where shareholders analyze infrastructure investments with large payouts and long-term gains.
This research is carried out by SMART and supported by the National Research Foundation (NRF) Singapore for its Research Complex and Technical Enterprise (CREATION) program.
The Future Urban Mobility Research Group uses new technological and institutional innovations to create the next generation of urban mobility systems to enhance access, equity, security and environmental performance for citizens and businesses in Singapore and other metropolitan areas worldwide. Could. FM is supported by NRF Singapore and is located at CREATE.
Smart was founded in 2007 by MIT in partnership with NRF Singapore. SMART serves as an intellectual and innovation hub for research relations between MIT and Singapore to undertake cutting-edge research projects in areas of interest for both Singapore and MIT. SMART currently consists of an innovation center and five interdisciplinary research groups: Antimicrobial Resistance, Manufacturing Personalized-Critical Analytics for Medicine, Agricultural Precision, Disruptive and Sustainable Technology for FM and Low Energy Electronic Systems.