Creating a new fragrance is a complex and time-consuming process that requires numerous trials and errors. To automate this process, researchers at the Tokyo Institute of Science have developed an artificial intelligence model called OGDiffusion, which creates new fragrances based on user-defined odor descriptors. The technology could be a game-changer for the perfume, food, and home goods industries.
Creating a new fragrance is a complex and time-consuming process that requires numerous trials and errors. To automate this process, researchers at the Tokyo Institute of Science have developed an artificial intelligence model called OGDiffusion, which creates new fragrances based on user-defined odor descriptors. The technology could be a game-changer for the perfume, food, and home goods industries.
OGDiffusion uses generative diffusion networks, a type of machine learning model that learns to create new content by modifying the noise process based on existing data. This technology, already widely used to create images and text, has been adapted by Japanese researchers to create new scents, Tech Xplore reports .
The system works by analyzing the chemical profiles (mass spectrometry data) of 166 essential oils, which are labeled with nine odor descriptors (e.g., "citrus" or "woody").
When users specify desired scent characteristics, the AI generates a corresponding chemical profile (mass spectrum) that matches those descriptors. It then calculates the blend of essential oils needed to recreate that scent using a mathematical method called least squares.
“Our diffusion network uses patterns in essential oil mass spectrometry data to create new aroma profiles using a fully automated, simplified, and data-driven approach, while maintaining high quality of the original data. By removing human intervention and molecular synthesis from the process, we offer a fast, versatile, and efficient method for creating aromas,” explains Professor Takamichi Nakamoto.
Existing AI-based fragrance creation models have been developed, but they rely on proprietary datasets and still require expert input. The main advantage of the new method is its ability to fully automate the creation of new fragrances. What’s more, because the system generates fragrances based on essential oil recipes, the final scent can be easily reproduced.
Even a beginner can create a conceived scent to create scented digital content, assures a Japanese professor.
Previously, dev.ua reported that an American company developed an AI Nose for robots .
