Activities

2.2. Development of platforms and artificial intelligence for sensory and neurosensory analysis of food products

Consumer acceptance is the biggest challenge when creating a new food product. There are several factors that determine acceptance, such as price, convenience and packaging, but one key factor that deserves significant evaluation is the sensory experience consumers have with the food. Sensory evaluation is the assessment of food products by trained individuals or consumers in terms of the human senses of sight, smell, taste, touch and hearing under controlled conditions. 

Technological advances in sensors have reduced their cost and size, allowing them to be used without inconveniencing users. In addition, recent advances in computational learning allow the implementation of Artificial Intelligence (AI) models capable of extracting complex knowledge from simpler sensory inputs, such as identifying dislike or aversion from facial expressions, sweat on the skin and changes in heart rate. 

Objectives:

1. Calibration of sensors and standardisation of the tasting process and questionnaires, using trained tasters; 
2. Collection of sensory and neurosensory test data in a controlled environment using trained tasters; 
3. Data pre-processing in order to create datasets for the subsequent training of AI models. Statistical analysis methodologies will be used for dimensionality reduction, parameter normalisation and feature engineering; 
4. Implementation, training and validation of AI models to identify the expressions and reactions of tasters to each of the different food products. Subsequent interactive process of fine tuning of the models by adjusting hyperparameters; 
5. Integration of AI models in the tasting data collection and analysis platform.

Lead: CATAA
Participants: IPN, CICYTEX, CTAEX, UEVORA