Cosmax develops AI scent prediction algorithm model..Prediction of scent based on molecular structure alone
Dec 03, 2025
Cosmax, a global cosmetics research, development, and production (ODM) company, will expand its use of AI to scent following usage and coloring. Cosmax expects AI-based fragrance prediction technology to improve the quality of cosmetics' olfactory sensibilities.
Cosmax announced on the 3rd that it has developed 'AI fragrance prediction algorithm model' that can predict fragrance only with molecular structure.
The AI scent prediction algorithm model is a technology that can predict olfactory characteristics for cosmetic raw materials including fragrances used in the cosmetics development process. This goes beyond the limitations of the existing method, which could only be confirmed by olfactory sensory evaluation, and can predict 'specificity' that can occur in cosmetics from the early stages of product development.
As cosmetics are developed by combining various raw materials, other raw materials in addition to fragrances have their own scent, so the results may have a different scent than expected. Since scents for cosmetic raw materials lack data and are mostly described only as specific odors, the process of identifying and modifying the problematic raw materials takes considerable time.
On the other hand, the AI fragrance prediction model developed by Cosmax is a model that can predict fragrance only with molecular structure by learning molecular fingerprints by machine learning based on vast molecular-fragrance data of 8,600 kinds. Among the used raw materials, the improvement work can be accelerated by identifying the raw materials that cause it.
In addition, this study has a high academic contribution to basic research on olfactory mechanisms. Cosmax published a paper on the study's achievements in the global scientific journal 『Communications Chemistry』, a sister journal of 『Nature』. In addition to cosmetics, it is expected that the scope of commercial application will be wide as it has potential for application in the fragrance, chemical, and food industries.
Cosmax announced on the 3rd that it has developed 'AI fragrance prediction algorithm model' that can predict fragrance only with molecular structure.
The AI scent prediction algorithm model is a technology that can predict olfactory characteristics for cosmetic raw materials including fragrances used in the cosmetics development process. This goes beyond the limitations of the existing method, which could only be confirmed by olfactory sensory evaluation, and can predict 'specificity' that can occur in cosmetics from the early stages of product development.
As cosmetics are developed by combining various raw materials, other raw materials in addition to fragrances have their own scent, so the results may have a different scent than expected. Since scents for cosmetic raw materials lack data and are mostly described only as specific odors, the process of identifying and modifying the problematic raw materials takes considerable time.
On the other hand, the AI fragrance prediction model developed by Cosmax is a model that can predict fragrance only with molecular structure by learning molecular fingerprints by machine learning based on vast molecular-fragrance data of 8,600 kinds. Among the used raw materials, the improvement work can be accelerated by identifying the raw materials that cause it.
In addition, this study has a high academic contribution to basic research on olfactory mechanisms. Cosmax published a paper on the study's achievements in the global scientific journal 『Communications Chemistry』, a sister journal of 『Nature』. In addition to cosmetics, it is expected that the scope of commercial application will be wide as it has potential for application in the fragrance, chemical, and food industries.
|
This article was translated by Naver AI translator.
