A Perception-Driven Framework for Predicting Missing Odor Perceptual Ratings and an Exploration of Odor Perceptual Space

Xin Li,Angus K Y Wong,Dehan Luo,Kevin Hung,Yu Cheng

doi:10.1109/access.2020.2972946

Xin Li, Angus K Y Wong + Show 3 more

Open Access

https://doi.org/10.1109/access.2020.2972946

Copy DOI

Abstract

There have been abundant research efforts on predicting verbal descriptions of odorants through the physicochemical features, physiological signals and E-nose signals. These approaches are interpreted as feature-driven methods in which the information about the inner links among different odor percepts is ignored. Different from that, we propose a perception-driven framework for predicting the missing odor perceptual ratings from other known odor percepts. Specifically, the work emphasizes pleasantness prediction based on level of importance in odor perception. In essence, this approach utilizes the relations among different odor perceptions, exploring the odor perceptual space subsequently. The missing perceptual ratings are predicted with an accuracy higher than 0.5 for more than half of the odor verbal descriptors, and almost half of the descriptors are predicted with a correlation higher than 0.8. The asymmetric clustering structure of odor perceptual space is revealed by feature selection for predicting the missing perceptual ratings. It is found that `pleasantness' is primarily determined by `sweet'.

Highlights

Compared with vision and audition, the structure of olfactory perceptual space is a more challenging issue, which stems primarily from the more complicated physiology mechanism of olfactory perception [1], [2]
The aim of this paper is to explore the relations among different odor percepts and reveal the structure of the odor perceptual space
Different from the previous feature-driven methods for predicting the target odor perceptual ratings [17]–[19], [22], [24]–[26], [30], [32]–[34], [36], [37], a perception-driven framework is proposed to predict the perceptual ratings of missing target odor percepts with a much better performance, and some rules of olfactory perception psychology are revealed to some extent, which could not be obtained though feature-driven methods

Summary

Introduction

Compared with vision and audition, the structure of olfactory perceptual space is a more challenging issue, which stems primarily from the more complicated physiology mechanism of olfactory perception [1], [2]. To address the problem of olfactory perception description, a set of domain-specific odor verbal descriptors have been developed to profile the odorants [8], [9], such as ‘sweet’, ‘fish’, ‘decayed’. These verbal descriptors, referred to as semantic attributes, can represent odorant perception reliably [10]. They constitute the odor perceptual space and can be considered as axises in the space [11].

Objectives

Methods

Results

Discussion

Conclusion