Interactive Machine Learning for Applications in Food Science

Tonda, Alberto; Boukhelifa, Nadia; Chabin, Thomas; Barnabe, Marc; Génot, Benoît; Lutton, Évelyne; Perrot, Nathalie

doi:10.1007/978-3-319-90403-0_22

Cited by 5 publications

(2 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Experts in a specific domain can provide invaluable insight into products and processes, but this precious knowledge is often available only in the form of intuition and non-coded expertise. Including expert insight in a model is not a straightforward process, but it can effectively be tackled by having humans interacting with a machine learning process, through visualization, or via specialists in encoding implicit domain knowledge [1][2][3].…”

Section: Introductionmentioning

confidence: 99%

Studying identification algorithms in food technology

Takhumova

Bondarenko

Shlykov

et al. 2019

Engineering for Rural Development

View full text Add to dashboard Cite

Food production specificity requires improvement of systems of automatic regulation of the processes in aggregates, apparatus, and facility. In creating an adaptive process control system for food production, based on the model of a control object, an additional analysis should be done to choose the identification algorithm of a real, fairly representative sample of input data and output measurement results. By now a lot of recurrent identification algorithms are development and in one or another extent investigated. The little part of these algorithms is implemented into software, such as Matlab, Unity Pro, and other similar ones. But the problem is that among all of these algorithms the best one for real conditions of identification is absent. And a practicing engineer, who takes up the implementation of an adaptive control system with an identifier, is faced with the problem of choosing a specific identification algorithm.In this research, by using simulation modelling, over 53 recurrent identification algorithms were analyzed, plus the main modifications of these algorithms, and a total of 47 possible estimation criteria for non-stationary multidimensional dynamic objects. Based on this analysis of the object class in question, several algorithms were recommended to engineering practice. Possibilities of a software package that has the most complete set of parametric identification algorithms today are discussed. For these specific conditions of the identification algorithms comparison into the package by identifying the stationary coefficients into object equation, the most effective were: Aizerman-1, Kachmazh, Nagumo-Noda, Rastrigin, Kalman Filter, Forgetting Factor, Tsypkin. With non-stationary coefficients of the object -Kachmazh, Nagumo-Noda, Kalman Filter. The best result was shown by the Nagumo-Noda identification algorithm.

show abstract

Section: Introductionmentioning

confidence: 99%

Studying identification algorithms in food technology

Takhumova

Bondarenko

Shlykov

et al. 2019

Engineering for Rural Development

View full text Add to dashboard Cite

show abstract

“…The use of machine learning (ML) has been applied to different industries such as sustainability of materials [ 36 ], techno-economics [ 37 ], molecular crystals engineering [ 38 ], energy [ 39 ], diagnostics in medicine [ 40 ] and, more recently, food/beverages [ 17 , 18 , 22 , 29 , 41 ] and agriculture [ 42 , 43 , 44 ]. This has been an effective tool to aid in the prediction and rapid assessment of products; however, a common issue found when using ML is the overfitting of the models because the generalization of the data is not achieved.…”

Section: Introductionmentioning

confidence: 99%

Integrating a Low-Cost Electronic Nose and Machine Learning Modelling to Assess Coffee Aroma Profile and Intensity

Viejo

Tongson

Fuentes

2021

Sensors

View full text Add to dashboard Cite

Aroma is one of the main attributes that consumers consider when appreciating and selecting a coffee; hence it is considered an important quality trait. However, the most common methods to assess aroma are based on expensive equipment or human senses through sensory evaluation, which is time-consuming and requires highly trained assessors to avoid subjectivity. Therefore, this study aimed to estimate the coffee intensity and aromas using a low-cost and portable electronic nose (e-nose) and machine learning modeling. For this purpose, triplicates of six commercial coffee samples with different intensity levels were used for this study. Two machine learning models were developed based on artificial neural networks using the data from the e-nose as inputs to (i) classify the samples into low, medium, and high-intensity (Model 1) and (ii) to predict the relative abundance of 45 different aromas (Model 2). Results showed that it is possible to estimate the intensity of coffees with high accuracy (98%; Model 1), as well as to predict the specific aromas obtaining a high correlation coefficient (R = 0.99), and no under- or over-fitting of the models were detected. The proposed contactless, nondestructive, rapid, reliable, and low-cost method showed to be effective in evaluating volatile compounds in coffee, which is a potential technique to be applied within all stages of the production process to detect any undesirable characteristics on–time and ensure high-quality products.

show abstract