Testing and enhancing spatial transferability of artificial neural networks based travel behavior models

ABSTRACT

Artificial Neural Networks (ANNs) are emerging classes of AI algorithms, and have seen numerous applications in travel behavior research recently. However, the transferability of ANN-based travel behavior models is seldom tested. A few studies that test transferability, merely use vanilla Feedforward Neural Networks. This paper evaluates the spatial transferability of two ANN-based models: first, a Feedforward ANN-based mode choice model, and next, a Long Short Term Memory (LSTM)-based activity generation and activity-timing model, and enhances their transferability using transfer learning (TL). Both the models were found to exhibit poor transferability in case of naïve transfer. Transfer learning resulted in significant improvements with the TL-enhanced models that utilizeonly 50% of local data achieving results similar to a locally developed model. Further, ANNs performed poorer when compared with nested logit (NL) models during naïve transfer. However, the TL-enhanced ANN-based models showed significant improvement compared to transfer scaling enhanced NL models.

KEYWORDS:

• Spatial transferability
• activity generation and timing model
• long short term memory networks
• Machine learning
• Transfer learning

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the FIRP Scheme, Indian Institute of Technology Delhi [MI02073].