Deep Design: Product Aesthetics for Heterogeneous Markets

Abstract

Aesthetic appeal is a primary driver of customer consideration for products such as automobiles. Product designers must accordingly convey design attributes (e.g., 'Sportiness'), a challenging proposition given the subjective nature of aesthetics and heterogeneous market segments with potentially different aesthetic preferences. We introduce a scalable deep learning approach that predicts how customers across different market segments perceive aesthetic designs and provides a visualization that can aid in product design. We tested this approach using a large-scale product design and crowdsourced customer data set with a Siamese neural network architecture containing a pair of conditional generative adversarial networks. The results show that the model predicts aesthetic design attributes of customers in heterogeneous market segments and provides a visualization of these aesthetic perceptions. This suggests that the proposed deep learning approach provides a scalable method for understanding customer aesthetic perceptions.