Machine learning for optimisation of flavours and fragrances
The design space that flavourists need to explore in optimising their chemicals, blends, and formulations can be vast. The complexity comes from a variety of sources: the huge range of potential ingredients and processing options; the need to scale-up processes; and from dozens of commercial, regulatory, and consumer constraints. The result can be a daunting amount of time-consuming experimentation. Using existing data from prior experiments and the literature can help. And many researchers are now turning to machine learning technologies, seeking to gain insights from this data that can provide an edge. But this brings its own challenges. The available data, collated from multiple sources, can often be sparse. It can also be noisy, particularly where biological processes or human sensory perceptions are in play, making exact control and reproducibility of experimental conditions difficult. And machine learning methodologies are typically very difficult to apply with sparse, noisy data.
In this webinar, we'll explore the potential of machine learning for flavourists. We'll discuss these challenges, and how they can be overcome. We'll introduce one machine learning method, Alchemite™, that is able to build machine learning models from sparse, noisy, experimental data. We'll show, with case studies, how this can be applied to guide experimental programmes, with typical reductions of 50-80% in time and cost, and to identify ingredient and process changes that lead to improved products. We'll discuss how machine learning increases the agility of R&D teams, enabling a faster response when design parameters change, for example, due to supply chain issues or regulations. The webinar will include a live demonstration of the Alchemite™ software and Q&A.
Dr Tom Whitehead
Tom is Head of Machine Learning at Intellegens. He has a PhD in theoretical physics from the University of Cambridge, and is now leading the Science Team that supports Intellegens customers in applying novel machine learning approaches to a wide variety of industrial applications. Through engagement in dozens of research projects, he has developed extensive experience in the practical application of machine learning to design experimental programmes, new materials and chemicals, formulated products, and manufacturing processes. Tom also leads the development of leading-edge machine learning and data analysis tools now used across multiple industrial R&D organisations.
Will is a sales & business development manager at Intellegens. He is supporting the adoption of Alchemite machine learning, with a particular focus on applying his experience in the food, flavour, and fragrance industries. He works closely with research teams to understand their specific requirements ensuring they gain maximum value from the technology.
Intellegens aims to be the leading machine learning solution for real-world, sparse and noisy data problems in industrial R&D and manufacturing processes. Its focus is on making it easy to apply machine learning to accelerate innovation. The Alchemite™ method originated at the University of Cambridge and development is on-going at Intellegens, in close collaboration with our growing community of Alchemite™ customer organisations. These represent sectors including alloys, additive manufacturing, aerospace, batteries, ceramics, chemical processes, composites, consumer products, cosmetics, drug discovery, energy, food and beverage, formulated products, paints, plastics, and printing technology.