Nathalie Perrot

Representing parametric probabilistic models tainted with imprecision

Numerical possibility theory, belief functions have been suggested as useful tools to represent imprecise, vague or incomplete information. They are particularly appropriate in uncertainty analysis where information is typically tainted with imprecision or incompleteness. Based on their experience or their knowledge about a random phenomenon, experts can sometimes provide a class of distributions without being able to precisely specify the parameters of a probability model. Frequentists use two-dimensional Monte-Carlo simulation to account for imprecision associated with the parameters of probability models. They hence hope to discover how variability and imprecision interact. This paper presents the limitations and disadvantages of this approach and propose a fuzzy random variable approach to treat this kind of knowledge.

Dry sausage ripening control integration of sensory-related properties

A feed-forward control algorithm is proposed to ensure the constancy of the sensory quality of the dry sausage during its fermentation. Fermentation is carried out with industrial equipment. The algorithm is based on human skill: the input variables of the controller are sensory evaluations made close to the line by the operators and the human diagnosis has been modeled using fuzzy logic. An experimental validation is introduced: the results show that the algorithm is likely to be able to control the process so as to obtain the desired sensory characteristics at the end of the fermentation stage. A user interface has been implemented in order to help the operator to cope with process control.

Modeling Human Expertise on a Cheese Ripening Industrial Process Using GP

Industrial agrifood processes often strongly rely on human expertise, expressed as know-how and control procedures based on subjective measurements (color, smell, texture), which are very difficult to capture and model. We deal in this paper with a cheese ripening process (of french Camembert), for which experimental data have been collected within a cheese ripening laboratory chain. A global and a monopopulation cooperative/coevolutive GP scheme (Parisian approach) have been developed in order to simulate phase prediction (i.e. a subjective estimation of human experts) from microbial proportions and Ph measurements. These two GP approaches are compared to Bayesian network modeling and simple multilinear learning algorithms. Preliminary results show the effectiveness and robustness of the Parisian GP approach.

Experimental analysis of sensory measurement imperfection impact for a cheese ripening fuzzy model

In the food processes, build tools taking human measurements into account is relevant for the control of the sensory quality of food products. Despite the methodology used to formalize these measurements, these ones are subjected to more imperfections (imprecision, reliability,...). Our aim is to develop tools taking these measurements into account and smoothing the imperfections of these measurements. In this paper, an experimental analysis is led on a fuzzy symbolic model applied to cheese ripening process. This analysis allows to determine the sensory measurements which have the highest impact on the model and to observe the impact of sensory measurements imperfection on the output of the developed fuzzy model.

Visual analysis of population scatterplots

We investigate how visual analytic tools can deal with the huge amount of data produced during the run of an evolutionary algorithm. We show, on toy examples and on two real life problems, how a multidimensional data visualisation tool like ScatterDice/GraphDice can be easily used for analysing raw output data produced along the run of an evolutionary algorithm. Visual interpretation of population data is not used very often by the EA community for experimental analysis. We show here that this approach may yield additional high level information that is hardly accessible through conventional computation.

A Dynamic Bayesian Network to Represent a Ripening Process of a Soft Mould Cheese

Available knowledge to describe food processes has been capitalized from different sources, is expressed under different forms and at different scales. To reconstruct the puzzle of knowledge by taking into account uncertainty, we need to combine, integrate different kinds of knowledge. Mathematical concepts such that expert systems, neural networks or mechanistic models reach operating limits. In all cases, we are faced with the limits of available data, mathematical formalism and the limits of human reasoning. Dynamical Bayesian Networks (DBNs) are practical probabilistic graphic models to represent dynamical complex systems tainted with uncertainty. This paper presents a simplified dynamic bayesian networks which allows to represent the dynamics of microorganisms in the ripening of a soft mould cheese (Camembert type) by means of an integrative sensory indicator. The aim is the understanding and modeling of the whole network of interacting entities taking place between the different levels of the process.

Estimation of the food product quality using fuzzy sets

The estimation of food product quality using fuzzy sets is discussed in this paper through two specific examples: (i) prediction of the luminance of biscuits during a baking process, and (ii) prediction of wet-milling quality of maize during a drying process. Two fuzzy approaches are validated: a black-box approach and a knowledge-based approach to modeling. The results are good and coherent in both cases and the models are robust. Nevertheless, the fuzzy knowledge-based modeling approach is particularly pertinent and adaptable to food process engineering research.

Editorial: Fuzzy concepts applied to food product quality control

Fuzzy logic is now a wide field of study and different tools have been developed over the last 10 years. Its implementation in food quality control for the food industry has been highlighted by sev...

Formalization of at-line Human Evaluations to Monitor Product Changes during Processing: The Concept of Sensory Indicator

Sensory characteristics of food products are essential for consumers. It is a challenge for firms to maintain these characteristics constant, with as few variations in quality as possible. The control of quality properties and in particular sensory ones can be carried out using reliable process control strategies. Nevertheless, classical control approaches can rarely be used in food processes due in particular to the lack of real-time, reliable instrumental sensors, which limits available information on the product, and to poor understanding of the interactions between food and process. The scarcity of suitable on-line sensors is closely related to the variability of the raw material, the complexity of the biological phenomena during processing and the severe constraints that sensors must satisfy, such as hygiene, high humidity, and so on. As a consequence, some food product properties are very difficult to be quantified during food manufacture [11]. However, various solutions have been explored to overcome this problem: sensor design and adaptation, off-line measurements, software sensors [21]. Moreover, human evaluation is widely accepted as a tool for the evaluation of the quality of food products: operators play a major role in process control, since they take into account not only the information from sensors but also that from their own senses [20]. They can detect small changes in product characteristics such as cookie color after baking [12] thanks to their process knowledge and experience.

Cooperative Coevolution for Agrifood Process Modeling

On the contrary to classical schemes of evolutionary optimisations algorithms, single population Cooperative Co-evolution techniques (CCEAs, also called “Parisian” approaches) make it possible to represent the evolved solution as an aggregation of several individuals (or even as a whole population). In other words, each individual represents only a part of the solution. This scheme allows simulating the principles of Darwinian evolution in a more economic way, which results in gain in robustness and efficiency. The counterpart however is a more complex design phase. In this chapter, we detail the design of efficient CCEAs schemes on two applications related to the modeling of an industrial agri-food process. The experiments correspond to complex optimisations encountered in the modeling of a Camembert-cheese ripening process. Two problems are considered: