It has been over two decades since LSRE initiated research in Perfume Engineering, a pioneering field aiming to develop scientific methodologies for the pre-formulation stages of perfume design, combining Chemical Engineering concepts with psychophysics for human sensory perception.

The team's outstanding contributions in this area are underscored by the publication of a key book: Perfume Engineering: Design, Performance & Classification. In 2004, this expertise led to the creation I-Sensis – Ideias Perfumadas, a spin-off company dedicated to personalized perfumes and olfactive marketing. In 2015, the doctoral thesis titled Perfume Performance and Classification: Perfumery Quaternary-Quinary Diagram (PQ2D®) and Perfumery Radar, received recognition as the best European doctoral thesis in Product Design & Engineering by the European Federation of Chemical Engineering (EFCE).

Research activities are targeted to the development of methodologies supporting the design, performance and classification of scented products, such as the Perfumery Ternary Diagram (PTD®) and Perfumery Quaternary-Quinary Diagram® (PQ2D®) methodologies, Diffusion model, Perfumery Radar (PR) and Perfumery Radar 2.0 (PR2.0) methodologies. The perfumery ternary diagram (PTD) methodology was a pioneering idea that can be extended to fragrance mixtures of N components to find compositions delivering a specific scent. It can be further elaborated to incorporate the effect of skin on the evaporation of perfumes. Perfumery Radar can be extended to other domains (e.g. wine industry), and the methodology can be extended to taste/flavour engineering.

The LSRE-LCM team has recently introduced an innovative approach, leveraging Deep Learning to assist in simulation-optimization processes. This study involved the identification of a surrogate model capable of learning the intricate dynamics of perfume release and propagation, correlating them with perception. The resulting framework introduced a pioneering paradigm for flavour engineering, seamlessly integrating web scraping, generative, and reinforcement learning within a transfer learning context. This complex system was designed to generate molecules with specific desired characteristics and chemical properties. These methodologies offer innovative approaches to generating molecules with specific characteristics and chemical properties, and obtaining optimal formulations, aligning with the ever-evolving demands of these dynamic industries. The proposed models and frameworks have shown promising results, indicating a transformative potential for product engineering in the perfume and flavour sectors.

Major projects in this research area include:

• Extraction and Identification of Natural Fragrances
• Perfume Design, Performance and Classification
• Perfumes and Flavors Using Scientific Machine Learning