Seokhee Jeon

From signals to Affection: Deep learning based affection estimation and signal authoring

In this talk, we introduce our lab’s “Signal to Affection” framework, which establishes a mapping between haptic signals and their corresponding descriptive adjectives. This association model enables accurate estimation of affective responses purely from haptic signals—allowing designers to evaluate the emotional characteristics of a target object without the need for a haptic rendering system. More practically, the framework also works in reverse: haptic signals can be synthesized based on input adjectives, providing an intuitive tool for designing affective haptic content.