# A Neural-embedded Choice Model: TasteNet-MNL Modeling Taste Heterogeneity with Flexibility and Interpretability

arXiv: https://arxiv.org/abs/2002.00922 · HTML: https://mlsanity.com/paper/2002.00922/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 3 Feb 2020.

## Built on by (most cited first, 22 of 22)

- [Interpretable Social Anchors for Human Trajectory Forecasting in Crowds](https://mlsanity.com/paper/2105.03136/) (2021)
- [Choice modelling in the age of machine learning -- discussion paper](https://mlsanity.com/paper/2101.11948/) (2021)
- [A Deep Generative Model for Feasible and Diverse Population Synthesis](https://mlsanity.com/paper/2208.01403/) (2022)
- [Aligning LLM with human travel choices: a persona-based embedding learning approach](https://mlsanity.com/paper/2505.19003/) (2025)
- [Auditing and Fixing Economic Validity in Tabular Foundation Models for Discrete Choice](https://mlsanity.com/paper/2605.26559/) (2026)
- [Deep neural networks for choice analysis: Enhancing behavioral regularity with gradient regularization](https://mlsanity.com/paper/2404.14701/) (2024)
- [Comparing hundreds of machine learning classifiers and discrete choice models in predicting travel behavior: an empirical benchmark](https://mlsanity.com/paper/2102.01130/) (2021)
- [Deep Learning for Choice Modeling](https://mlsanity.com/paper/2208.09325/) (2022)
- [RUMBoost: Gradient Boosted Random Utility Models](https://mlsanity.com/paper/2401.11954/) (2024)
- [Designing Graph Convolutional Neural Networks for Discrete Choice with Network Effects](https://mlsanity.com/paper/2503.09786/) (2025)
- [Embedding Foundation Model Predictions in Discrete-Choice Models with Structural Guarantees](https://mlsanity.com/paper/2606.26432/) (2026)
- [Incorporating Domain Knowledge in Deep Neural Networks for Discrete Choice Models](https://mlsanity.com/paper/2306.00016/) (2023)
- [Choice Models and Permutation Invariance: Demand Estimation in Differentiated Products Markets](https://mlsanity.com/paper/2307.07090/) (2023)
- [Images in Discrete Choice Modeling: Addressing Data Isomorphism in Multi-Modality Inputs](https://mlsanity.com/paper/2312.14724/) (2023)
- [Fully Data-driven but Interpretable Human Behavioural Modelling with Differentiable Discrete Choice Model](https://mlsanity.com/paper/2412.19403/) (2024)
- [Context-aware Bayesian Mixed Multinomial Logit Model](https://mlsanity.com/paper/2210.05737/) (2022)
- [Bayesian Deep Learning for Discrete Choice](https://mlsanity.com/paper/2505.18077/) (2025)
- [Graph neural networks for residential location choice: connection to classical logit models](https://mlsanity.com/paper/2507.21334/) (2025)
- [DeepLogit: A sequentially constrained explainable deep learning modeling approach for transport policy analysis](https://mlsanity.com/paper/2509.13633/) (2025)
- [Neural-Bayesian Structure Learning for Discrete Choice Modeling](https://mlsanity.com/paper/2608.25258/) (2026)
- [Consideration Circuits: Depth Separation and Universality Beyond a Single Softmax](https://mlsanity.com/paper/2610.04143/) (2026)
- [Better Capturing Interactions Between Products in~Retail: Revisited Negative Sampling for~Basket Choice Modeling](https://mlsanity.com/paper/2610.17963/) (2026)

## Abstract (arXiv)

Discrete choice models (DCMs) require a priori knowledge of the utility functions, especially how tastes vary across individuals. Utility misspecification may lead to biased estimates, inaccurate interpretations and limited predictability. In this paper, we utilize a neural network to learn taste representation. Our formulation consists of two modules: a neural network (TasteNet) that learns taste parameters (e.g., time coefficient) as flexible functions of individual characteristics; and a multinomial logit (MNL) model with utility functions defined with expert knowledge. Taste parameters learned by the neural network are fed into the choice model and link the two modules. Our approach extends the L-MNL model (Sifringer et al., 2020) by allowing the neural network to learn the interactions between individual characteristics and alternative attributes. Moreover, we formalize and strengthen the interpretability condition - requiring realistic estimates of behavior indicators (e.g., value-of-time, elasticity) at the disaggregated level, which is crucial for a model to be suitable for scenario analysis and policy decisions. Through a unique network architecture and parameter transformation, we incorporate prior knowledge and guide the neural network to output realistic behavior indicators at the disaggregated level. We show that TasteNet-MNL reaches the ground-truth model's predictability and recovers the nonlinear taste functions on synthetic data. Its estimated value-of-time and choice elasticities at the individual level are close to the ground truth. On a publicly available Swissmetro dataset, TasteNet-MNL outperforms benchmarking MNLs and Mixed Logit model's predictability. It learns a broader spectrum of taste variations within the population and suggests a higher average value-of-time.