Acessibilidade / Reportar erro

Application of neural networks in predicting the qualitative characteristics of fruits

Abstract

In this research, the quality properties of persimmon were predicted using artificial intellect techniques. The persimmon samples were transferred to a computer vision lab, room temperature of 24 °C and 22% RH. The samples were divided into three groups for temperature treatment. They were kept at three temperature levels of 5 °C, 15 °C, and 24°C (control group) for 72 hours. The sample was then placed at room temperature and was imaged every second day for a 14 day period. After imaging, each sample underwent destructive tests to determine their quality attributes, including sugar content, firmness, and pH. The results indicate that the neural network's predicted values of acidity, firmness, and sugar of persimmon were not statistically significant differences from their actual values. In predicting the acidity of persimmon, the sugar RMSE is more than the two factors of firmness and acidity. For this reason, the accuracy of firmness and acidity is higher than sugar. MAPE is 10.11, 20.81, and 6.03 for acidity, firmness, and sugar, respectively. The model for sugar indicates a high difference between the actual values and the predicted values.

Keywords:
artificial neural networks; firmness; persimmon; acidity; sugar

1 Introduction

Recently, consumer demand for high-quality fruits and vegetables has been increasing. Food quality and health is important factor in the modern food industry. Due to this, increasing the accuracy and reliability of the post-harvest operations, including quality assessment, should be done for fruits. Conventional food evaluation methods are now destructive and inefficient. Therefore, it is important to develop a non-destructive evaluation tool.

One of the most important techniques in medicine for diagnosing diseases is Magnetic Resonance Imaging (MRI), which can be used to take clear scans of various tissues of the body without X-rays. MRI is suitable for non-destructive imaging of water-containing materials and is widely used in various fields, especially food science and industry (Rogerio et al., 2014Rogerio, W. F., Greiner, R., Nunes, I. L., Feitosa, S., Furtunato, D. M. N., & Almeida, D. T. (2014). Effect of preparation practices and the cowpea cultivar Vigna unguiculata L.Walp on the quality and content of myo-inositol phosphate in akara (fried bean paste). Food Science and Technology, 34(2), 243-248. http://dx.doi.org/10.1590/fst.2014.0040.
http://dx.doi.org/10.1590/fst.2014.0040...
). NMR stands for Nuclear Magnetic Resonance. The researchers were able to show the absorption of electromagnetic radiation as a result of the transfer of the energy level of the nucleus in a strong magnetic field (Silva et al., 2019Silva, L. A., Flumignan, D. L., Tininis, A. G., Pezza, H. R., & Pezza, L. (2019). Discrimination of Brazilian lager beer by 1H NMR spectroscopy combined with chemometrics. Food Chemistry, 272, 488-493. http://dx.doi.org/10.1016/j.foodchem.2018.08.077. PMid:30309572.
http://dx.doi.org/10.1016/j.foodchem.201...
). Like H +, which have magnetic properties, they act like small magnets. These particles when placed in a magnetic field. They are located in the direction of the field. Also, these particles change direction under the influence of radio waves, and when the current radio source is cut off, these waves begin to move to their original place (Luiz & Vanin, 2021Luiz, R. O., & Vanin, F. M. (2021). Effect of straight dough X pre-fermented dough method on composite wheat breads characteristics. Food Science and Technology. Ahead of Print. http://dx.doi.org/10.1590/fst.64420.
http://dx.doi.org/10.1590/fst.64420...
). The speed of movement depends on the connection of this valley to other elements instead of the original. If this valley has a strong connection with other elements, the speed of return to the original state will be fast, and if it does not have a strong connection with other elements, the speed of return to the original state will be slow (Hamdan et al., 2021Hamdan, A. N. A., Al-Saad, Z. A. A., & Abu-Alhail, S. (2021). Fuzzy system modelling to assess water quality for irrigation purposes. Journal of Water and Land Development, 50, 98-107.). In fact, in this system, the physicochemical properties of the material can be understood from the initial velocity of the valley.

One of the most important applications of image processing is as a non-destructive method to examine the texture of fruits during storage and storage. Due to the studies done with this technique, the inside of the fruits can also be examined (Ali, 2017Ali, S. S. (2017). Review paper on efficient quality inspection of food products using neural network classification. Journal of Artificial Intelligence Research & Advances, 4(1), 1-14.). For example, in pomegranate fruit, the damaged and damaged parts of the fruit can be determined by preparing this concentrate before preparing the concentrate and using the cutting steps. Separate those parts from pomegranate. Other applications of this method in modern agriculture are the study of the ripening stage of fruits, the study of the invasion of pathogens, tissue characteristics, transmission and diffusion, and the release of oxygen in agricultural products (Nayak et al., 2020Nayak, J., Vakula, K., Dinesh, P., Naik, P., & Pelusi, D. (2020). Intelligent food processing: Journey from artificial neural network to deep learning. Computer Science Review, 38, 100297. http://dx.doi.org/10.1016/j.cosrev.2020.100297.
http://dx.doi.org/10.1016/j.cosrev.2020....
). This non-destructive technique can study the distribution and motility of protons in water molecules and other metabolites concentrated in biological tissue. This method makes it possible to determine the change in the concentration of oil and water in food and agricultural products, which is usually associated with fruit blight, damage, and rot. By determining this factor, it is easy to understand the maturity or spoilage of the fruit. Some researchers have shown rapid transfer of water in grains during their studies on corn using image processing (Oliveira et al., 2021Oliveira, T. F., Clemente, M. A. J., Galvan, D., Fix, G., Mantovani, A. C. G., Moreira, I., Andrello, A. C., Angilelli, K. B., & Borsato, D. (2021). Application of artificial neural networks in the study of Mozzarella cheese salting. Food Science and Technology, 41(Suppl. 1), 375-385. http://dx.doi.org/10.1590/fst.18320.
http://dx.doi.org/10.1590/fst.18320...
).

Moisture is usually removed from the crop during drying, which is a very important amount of moisture left in agricultural products, so that if a little moisture is removed from the crop due to drying, it may cause mold during storage or if too much moisture is removed. Cause wrinkles and tissue damage. In rice, the drying stage is also very important, so that if the rice dries properly, it will find the best quality during cooking. Image processing is a good way to check the drying process of agricultural products. For example, in a study on drying rice, the results showed that drying the amount of water up to 60 °C increases rapidly, but then a slight decrease in the amount of water can be seen (Parewai et al., 2020Parewai, I., As, M., Mine, T., & Koeppen, M. (2020). Identification and classification of sashimi food using multispectral technology. In Proceedings of the 2020 2nd Asia Pacific Information Technology Conference (pp. 66-72.). New York: ACM. http://dx.doi.org/10.1145/3379310.3379317.
http://dx.doi.org/10.1145/3379310.337931...
).

Freezing is one of the most important methods of storing vegetables, especially for a long time. To re-use frozen vegetables, the defrosting process is necessary. This freezing process causes more damage to the plant tissue (Sun et al., 2019Sun, Q., Zhang, M., & Mujumdar, A. S. (2019). Recent developments of artificial intelligence in drying of fresh food: a review. Critical Reviews in Food Science and Nutrition, 59(14), 2258-2275. http://dx.doi.org/10.1080/10408398.2018.1446900. PMid:29493285.
http://dx.doi.org/10.1080/10408398.2018....
). Using the MRI technique, the best temperature and the best time for the freezing and thawing process can be determined using imaging of vegetables and fruits (Zhu et al., 2021Zhu, L., Spachos, P., Pensini, E., & Plataniotis, K. N. (2021). Deep learning and machine vision for food processing: a survey. Current Research in Food Science, 4, 233-249. http://dx.doi.org/10.1016/j.crfs.2021.03.009. PMid:33937871.
http://dx.doi.org/10.1016/j.crfs.2021.03...
). For example, in a study of okra, asparagus, soybeans, and broad beans, the results showed that during the thawing process, the curves from the MRI signals in okra and asparagus were linear, but for soybeans and broad beans, they were convex and convex (Djuris et al., 2020Djuris, J., Vidovic, V., & Ibric, S. (2020). Chapter nine: Release modeling of nanoencapsulated food ingredients by artificial intelligence algorithms. In S. M. Jafari (Ed.), Release and bioavailability of nanoencapsulated food ingredients (Vol. 5, pp. 311-347). Cambridge: Academic Press. http://dx.doi.org/10.1016/B978-0-12-815665-0.00009-6.
http://dx.doi.org/10.1016/B978-0-12-8156...
).

Numerous studies have been performed on the process of freezing and thawing meat using the MRI technique. This technique can be used to study the process of cooking meat to obtain the best quality of meat in terms of sensory evaluation. This method can be used to determine the interaction of sharp broth. With this method, the effect of different freezing methods on salmon muscles can be seen (Bhagya-Raj & Dash, 2020Bhagya-Raj, G. V. S., & Dash, K. K. (2020). Comprehensive study on applications of artificial neural network in food process modeling. Critical Reviews in Food Science and Nutrition. Ahead of print. PMid:33327740.).

There are several definitions for artificial intelligence, all of which can be placed in the two main approaches of weak artificial intelligence and strong artificial intelligence. A strong approach to the problem of artificial intelligence is building a machine with all the capabilities associated with intelligence in humans (Xu et al., 2021Xu, X., Ren, S., Wang, D., Ma, J., Yan, X., Guo, Y., Liu, X., & Pan, Y. (2021). Optimization of extraction of defatted walnut powder by ultrasonic assisted and artificical neural network. Food Science and Technology. Ahead of Print.http://dx.doi.org/10.1590/fst.53320.
http://dx.doi.org/10.1590/fst.53320...
). Nowadays, due to the expansion and more complex decision-making, information systems, especially industrial intelligence systems, have become more important in supporting decision-making (Khadir, 2021Khadir, M. T. (2021). Artificial neural networks in food processing: modelling and predictive control. Berlin: De Gruyter. http://dx.doi.org/10.1515/9783110646054.
http://dx.doi.org/10.1515/9783110646054...
). Artificial intelligence, then, refers to systems that can behave similarly to intelligent human behaviors (such as understanding complex situations, simulating human thought processes and reasoning methods, and learning and the ability to know and reason to solve problems) (Das et al., 2021Das, M., Alic, E., & Akpinar, E. K. (2021). Detailed analysis of mass transfer in solar food dryer with different methods. International Communications in Heat and Mass Transfer, 128, 105600. http://dx.doi.org/10.1016/j.icheatmasstransfer.2021.105600.
http://dx.doi.org/10.1016/j.icheatmasstr...
).

Expert systems are grounded in applications in artificial intelligence and engineering, which due to the growing need of societies to adopt solutions and quick decisions in cases where complex and multiple human knowledge is needed, their importance is increased. Expert systems solve problems that typically require expertise and expertise (Chen & Yu, 2021Chen, T. C., & Yu, S. Y. (2021). The review of food safety inspection system based on artificial intelligence, image processing, and robotic. Food Science and Technology. Ahead of print. http://dx.doi.org/10.1590/fst.35421.
http://dx.doi.org/10.1590/fst.35421...
). Machine iodine is known as expert system iodine only if it has a series of special capabilities, such as knowing its existence, which means that the machine is aware of its existence (Alamir, 2021Alamir, M. A. (2021). An enhanced artificial neural network model using the Harris Hawks optimiser for predicting food liking in the presence of background noise. Applied Acoustics, 178, 108022. http://dx.doi.org/10.1016/j.apacoust.2021.108022.
http://dx.doi.org/10.1016/j.apacoust.202...
).

In a communication network, communication signals are in the form of electrical waves. It is the main component of neurons, which consists of a cellular structure and a set of grooves and lines, and the grooves are where the information enters the neurons and the lines where the information leaves the neurons (Gonçalves et al., 2021Gonçalves, J. No., Ozorio, L. V., Abreu, T. C. C., Santos, B. F., & Pradelle, F. (2021). Modelling of biogas production from food, fruits and vegetables wastes using artificial neural network (ANN). Fuel, 281, 119081. http://dx.doi.org/10.1016/j.fuel.2020.119081.
http://dx.doi.org/10.1016/j.fuel.2020.11...
). The connection of the neuron to the other neuron is calledsenations, which acts as a gateway or key. If the reactions that millions of different neurons give to different dulcimers coincide with each other, important observations may occur in the brain (Jawa et al., 2020Jawa, C. R., Jumaat, S. A., & Yahya, M. (2020). Development of Artificial Neural Network (ANN)-based model to predict electricity production from food waste biogas system. In K. Isa, Z. M. Zain, R. Mohd-Mokhtar, M. M. Noh, Z. H. Ismail, A. A. Yusof, A. F. M. Ayob, S. S. A. Ali & H. A. Kadir (Eds.), Proceedings of the 12th National Technical Seminar on Unmanned System Technology (pp. 1081-1089). Singapore: Springer.). Of all the AI disciplines, the most useful are computerized and mechanized visual systems. The range of applications of the growing technology is very wide and ranges from common applications such as production line quality control and video surveillance to new technologies such as driverless cars. The range of applications of technology varies based on the techniques used in them (Ali & Dildar, 2021Ali, S. S. E., & Dildar, S. A. (2021). MLP-WOA neural network-based automated grading of fruits and vegetable quality detection for food industry using artificial intelligence techniques (Computer vision – image recognition). In X. Z. Gao, R. Kumar, S. Srivastava & B. P. Soni (Eds.), Applications of artificial intelligence in engineering (pp. 539-561). Singapore: Springer. http://dx.doi.org/10.1007/978-981-33-4604-8_43.
http://dx.doi.org/10.1007/978-981-33-460...
).

Some of our visual object systems, capable of shooting in the visible range, are also able to market objects in the invisible range. The information received from objects in the range of colored light can be useful in determining the maturity of plants, disease, and body and determining varieties, mating, the composition of functional properties, and contamination and disease of plants, seeds, fruits, vegetables, and fruits (Nosratabadi et al., 2021Nosratabadi, S., Ardabili, S., Lakner, Z., Mako, C., & Mosavi, A. (2021). Prediction of food production using machine learning algorithms of multilayer perceptron and ANFIS. Agriculture, 11(5), 408. http://dx.doi.org/10.3390/agriculture11050408.
http://dx.doi.org/10.3390/agriculture110...
).

2 Material and methods

The persimmon samples were transferred to a computer vision lab, room temperature of 24 °C and 22% RH. The samples were divided into three groups for temperature treatment. They were kept at three temperature levels of 5 °C, 15 °C, and 24 °C (control group) for 72 hours. The sample was then placed at room temperature and was imaged every second day for a 14 day period. After imaging, each sample underwent destructive tests to determine their quality attributes, including sugar content, firmness, and acidity.

2.1 Imaging

A 32-megapixel camera captured images in the visible spectrum. In order to fully position the samples in the field of view of the camera, according to the focal length of the lens in the camera, the distance between the camera and the samples was considered to be 20 cm. Imaging is done in such a way that the samples are placed under the camera, and the image is taken from a specified view of the sample. Proper lighting during shooting makes it easy to analyze images. Lighting is such as to prevent the formation of shadows and light focal points in the image. A blue background was also used for imaging. A refractometer was used to measure fruit sugar.

2.2 Artificial neural networks

In a communication network, communication signals are in the form of electrical waves. It is the main component of neurons which consists of a cellular structure and a set of grooves and lines. In this paper, a multilayer network of perceptron with a hidden layer has been used. The optimum number of neurons in the hidden layer was found to be 25. The maximum iterations were set as 1000, and the learning rate was set as 0.001. Input variables include temperature treatment at three levels (control, 15 and 5 °C), color channels (L, a, b), and the standard deviation of color channels (stdl, stda, stdb). Output variables include sugar, acidity, and firmness. The next step after preparing the data is to normalize it so that the training and test data sets have a statistically almost uniform distribution. This conversion must also be done in the range of appropriate changes to converge the network to the optimal point. Therefore, the best data conversion domain is 0.1 to 0.9 for the Tansig network (Rohani et al., 2011Rohani, A., Abbaspour-Fard, M. H., & Abdolahpour, S. (2011). Prediction of tractor repair and maintenance costs using Artificial Neural Network. Expert Systems with Applications, 38(7), 8999-9007. http://dx.doi.org/10.1016/j.eswa.2011.01.118.
http://dx.doi.org/10.1016/j.eswa.2011.01...
). For this purpose, linear normalization has been used to convert the data (Equation 1) (Nourani et al., 2019Nourani, V., Molajou, A., Najafi, H., & Mehr, A. D. (2019). Emotional ANN (EANN): a new generation of neural networks for hydrological modeling in IoT. In F. Al-Turjman (Ed.), Artificial intelligence in IoT (pp. 45-61). Cham: Springer. http://dx.doi.org/10.1007/978-3-030-04110-6_3.
http://dx.doi.org/10.1007/978-3-030-0411...
; Molajou et al., 2021Molajou, A., Nourani, V., Afshar, A., Khosravi, M., & Brysiewicz, A. (2021). Optimal design and feature selection by genetic algorithm for Emotional Artificial Neural Network (EANN) in rainfall-runoff modeling. Water Resources Management, 35(8), 2369-2384. http://dx.doi.org/10.1007/s11269-021-02818-2.
http://dx.doi.org/10.1007/s11269-021-028...
):

x n = x x m i n x m a x x m i n × r m a x r min + r min (1)

Where x is primary data, xn is normalized data, xmax and xmin are maximum and minimum of primary data, respectively, and rmax and rmin are maximum and minimum of conversion data, respectively. 70% of the data were used as training data, 15% of the data were used for the model validation phase, and the remaining 15% in the test phase.

3 Results and discussion

Figure 1 shows the neural network performance diagram for the three modes of training, validation, and testing for all three modes of acidity, stiffness, and sugar. Figure 1 includes three blue, green, and red lines for training, validation, and testing data, respectively. In this figure, the horizontal axis indicates the number of repetitions of the network training in which the network repetition is trained as can be seen for the factor of acidity the number of repetitions of the network respectively equal to 6 (mean square error value is 0.055). Meanwhile, the number of repetitions of the network for firmness is 7 (mean square error value is 0.018) and for sugar is 8 (mean square error value is 0.007). After this stage, the error of the training, validation, and testing steps has been constant. Functional diagrams start from a point with a specific error, and by repeating the network training, the error value is reduced and fixed after the convergence point. As can be seen, to predict the acidity of persimmon, the squares error is greater than the two factors of firmness and sugar.

Figure 1
Neural network performance diagram for three acidity variables: a) Acidity, b) Firmness, and c) Sugar.

Table 1 shows the evaluation factors of the neural network model for the three variables of acidity, firmness, and sugar, respectively. The RMSE (Root Mean Square Error) values are 0.69, 0.72, and 0.77 for acidity, firmness, and sugar. Furthermore, MAPE indicates the average neural network prediction error. MAPE is 10.11, 20.81, and 6.03 for acidity, firmness, and sugar, respectively. According to the results, the network Efficiency Factor (EF) for firmness is greater than acidity and sugar.

Table 1
Evaluation factors of neural network model related to output variables.

Table 2 presents linear regression models in three training, validation, and test phases. According to Table 2, in the training and validation phase, the obtained relationships have a high coefficient of explanation. In the test phase of the obtained models, the value of the model explanation coefficient for prediction of acidity is 0.83, while this coefficient is 0.89 and 0.30 for prediction of firmness and sugar. The results show that the obtained relations are accurate enough to predict the acidity and firmness factors. However, prediction of sugar using this method is possible with non-applied accuracy.

Table 2
Predictive regression models from MLP neural network for three variables of acidity, firmness, and sugar.

4 Conclusion

In this study, the performance of the MLP artificial neural network was predicted to evaluate the acidity, firmness, and sugar of persimmon. The results indicate that the neural network's predicted values of acidity, firmness, and sugar of persimmon were not statistically significant differences from their actual values. In predicting the acidity of persimmon, the sugar RMSE is more than the two factors of firmness and acidity. For this reason, the accuracy of firmness and acidity is higher than sugar. MAPE is 10.11, 20.81, and 6.03 for acidity, firmness, and sugar, respectively. Comparison of the performance criteria of the neural network model showed that the firmness and acidity models are very close to each other, whereas the neural network model for sugar indicates a high difference between the actual values and the predicted values. Comparison of neural network results in training and testing stages showed that this technique could be used as a reliable method for estimating firmness and acidity factors with sufficient accuracy.

Acknowledgment

This research was funded by Princess Nourah bint Abdulrahman University Researchers Supporting Project number (PNURSP2022R145), Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia.

  • Practical Application: The neural network's predicted values of acidity, firmness, and sugar of persimmon were not statistically significant differences from their actual values. In predicting the acidity of persimmon, the sugar RMSE is more than the two factors of firmness and acidity.

References

  • Alamir, M. A. (2021). An enhanced artificial neural network model using the Harris Hawks optimiser for predicting food liking in the presence of background noise. Applied Acoustics, 178, 108022. http://dx.doi.org/10.1016/j.apacoust.2021.108022
    » http://dx.doi.org/10.1016/j.apacoust.2021.108022
  • Ali, S. S. (2017). Review paper on efficient quality inspection of food products using neural network classification. Journal of Artificial Intelligence Research & Advances, 4(1), 1-14.
  • Ali, S. S. E., & Dildar, S. A. (2021). MLP-WOA neural network-based automated grading of fruits and vegetable quality detection for food industry using artificial intelligence techniques (Computer vision – image recognition). In X. Z. Gao, R. Kumar, S. Srivastava & B. P. Soni (Eds.), Applications of artificial intelligence in engineering (pp. 539-561). Singapore: Springer. http://dx.doi.org/10.1007/978-981-33-4604-8_43
    » http://dx.doi.org/10.1007/978-981-33-4604-8_43
  • Bhagya-Raj, G. V. S., & Dash, K. K. (2020). Comprehensive study on applications of artificial neural network in food process modeling. Critical Reviews in Food Science and Nutrition Ahead of print. PMid:33327740.
  • Chen, T. C., & Yu, S. Y. (2021). The review of food safety inspection system based on artificial intelligence, image processing, and robotic. Food Science and Technology Ahead of print. http://dx.doi.org/10.1590/fst.35421
    » http://dx.doi.org/10.1590/fst.35421
  • Das, M., Alic, E., & Akpinar, E. K. (2021). Detailed analysis of mass transfer in solar food dryer with different methods. International Communications in Heat and Mass Transfer, 128, 105600. http://dx.doi.org/10.1016/j.icheatmasstransfer.2021.105600
    » http://dx.doi.org/10.1016/j.icheatmasstransfer.2021.105600
  • Djuris, J., Vidovic, V., & Ibric, S. (2020). Chapter nine: Release modeling of nanoencapsulated food ingredients by artificial intelligence algorithms. In S. M. Jafari (Ed.), Release and bioavailability of nanoencapsulated food ingredients (Vol. 5, pp. 311-347). Cambridge: Academic Press. http://dx.doi.org/10.1016/B978-0-12-815665-0.00009-6
    » http://dx.doi.org/10.1016/B978-0-12-815665-0.00009-6
  • Gonçalves, J. No., Ozorio, L. V., Abreu, T. C. C., Santos, B. F., & Pradelle, F. (2021). Modelling of biogas production from food, fruits and vegetables wastes using artificial neural network (ANN). Fuel, 281, 119081. http://dx.doi.org/10.1016/j.fuel.2020.119081
    » http://dx.doi.org/10.1016/j.fuel.2020.119081
  • Hamdan, A. N. A., Al-Saad, Z. A. A., & Abu-Alhail, S. (2021). Fuzzy system modelling to assess water quality for irrigation purposes. Journal of Water and Land Development, 50, 98-107.
  • Jawa, C. R., Jumaat, S. A., & Yahya, M. (2020). Development of Artificial Neural Network (ANN)-based model to predict electricity production from food waste biogas system. In K. Isa, Z. M. Zain, R. Mohd-Mokhtar, M. M. Noh, Z. H. Ismail, A. A. Yusof, A. F. M. Ayob, S. S. A. Ali & H. A. Kadir (Eds.), Proceedings of the 12th National Technical Seminar on Unmanned System Technology (pp. 1081-1089). Singapore: Springer.
  • Khadir, M. T. (2021). Artificial neural networks in food processing: modelling and predictive control Berlin: De Gruyter. http://dx.doi.org/10.1515/9783110646054
    » http://dx.doi.org/10.1515/9783110646054
  • Luiz, R. O., & Vanin, F. M. (2021). Effect of straight dough X pre-fermented dough method on composite wheat breads characteristics. Food Science and Technology Ahead of Print. http://dx.doi.org/10.1590/fst.64420
    » http://dx.doi.org/10.1590/fst.64420
  • Molajou, A., Nourani, V., Afshar, A., Khosravi, M., & Brysiewicz, A. (2021). Optimal design and feature selection by genetic algorithm for Emotional Artificial Neural Network (EANN) in rainfall-runoff modeling. Water Resources Management, 35(8), 2369-2384. http://dx.doi.org/10.1007/s11269-021-02818-2
    » http://dx.doi.org/10.1007/s11269-021-02818-2
  • Nayak, J., Vakula, K., Dinesh, P., Naik, P., & Pelusi, D. (2020). Intelligent food processing: Journey from artificial neural network to deep learning. Computer Science Review, 38, 100297. http://dx.doi.org/10.1016/j.cosrev.2020.100297
    » http://dx.doi.org/10.1016/j.cosrev.2020.100297
  • Nosratabadi, S., Ardabili, S., Lakner, Z., Mako, C., & Mosavi, A. (2021). Prediction of food production using machine learning algorithms of multilayer perceptron and ANFIS. Agriculture, 11(5), 408. http://dx.doi.org/10.3390/agriculture11050408
    » http://dx.doi.org/10.3390/agriculture11050408
  • Nourani, V., Molajou, A., Najafi, H., & Mehr, A. D. (2019). Emotional ANN (EANN): a new generation of neural networks for hydrological modeling in IoT. In F. Al-Turjman (Ed.), Artificial intelligence in IoT (pp. 45-61). Cham: Springer. http://dx.doi.org/10.1007/978-3-030-04110-6_3
    » http://dx.doi.org/10.1007/978-3-030-04110-6_3
  • Oliveira, T. F., Clemente, M. A. J., Galvan, D., Fix, G., Mantovani, A. C. G., Moreira, I., Andrello, A. C., Angilelli, K. B., & Borsato, D. (2021). Application of artificial neural networks in the study of Mozzarella cheese salting. Food Science and Technology, 41(Suppl. 1), 375-385. http://dx.doi.org/10.1590/fst.18320
    » http://dx.doi.org/10.1590/fst.18320
  • Parewai, I., As, M., Mine, T., & Koeppen, M. (2020). Identification and classification of sashimi food using multispectral technology. In Proceedings of the 2020 2nd Asia Pacific Information Technology Conference (pp. 66-72.). New York: ACM. http://dx.doi.org/10.1145/3379310.3379317
    » http://dx.doi.org/10.1145/3379310.3379317
  • Rogerio, W. F., Greiner, R., Nunes, I. L., Feitosa, S., Furtunato, D. M. N., & Almeida, D. T. (2014). Effect of preparation practices and the cowpea cultivar Vigna unguiculata L.Walp on the quality and content of myo-inositol phosphate in akara (fried bean paste). Food Science and Technology, 34(2), 243-248. http://dx.doi.org/10.1590/fst.2014.0040
    » http://dx.doi.org/10.1590/fst.2014.0040
  • Rohani, A., Abbaspour-Fard, M. H., & Abdolahpour, S. (2011). Prediction of tractor repair and maintenance costs using Artificial Neural Network. Expert Systems with Applications, 38(7), 8999-9007. http://dx.doi.org/10.1016/j.eswa.2011.01.118
    » http://dx.doi.org/10.1016/j.eswa.2011.01.118
  • Silva, L. A., Flumignan, D. L., Tininis, A. G., Pezza, H. R., & Pezza, L. (2019). Discrimination of Brazilian lager beer by 1H NMR spectroscopy combined with chemometrics. Food Chemistry, 272, 488-493. http://dx.doi.org/10.1016/j.foodchem.2018.08.077 PMid:30309572.
    » http://dx.doi.org/10.1016/j.foodchem.2018.08.077
  • Sun, Q., Zhang, M., & Mujumdar, A. S. (2019). Recent developments of artificial intelligence in drying of fresh food: a review. Critical Reviews in Food Science and Nutrition, 59(14), 2258-2275. http://dx.doi.org/10.1080/10408398.2018.1446900 PMid:29493285.
    » http://dx.doi.org/10.1080/10408398.2018.1446900
  • Xu, X., Ren, S., Wang, D., Ma, J., Yan, X., Guo, Y., Liu, X., & Pan, Y. (2021). Optimization of extraction of defatted walnut powder by ultrasonic assisted and artificical neural network. Food Science and Technology Ahead of Print.http://dx.doi.org/10.1590/fst.53320
    » http://dx.doi.org/10.1590/fst.53320
  • Zhu, L., Spachos, P., Pensini, E., & Plataniotis, K. N. (2021). Deep learning and machine vision for food processing: a survey. Current Research in Food Science, 4, 233-249. http://dx.doi.org/10.1016/j.crfs.2021.03.009 PMid:33937871.
    » http://dx.doi.org/10.1016/j.crfs.2021.03.009

Publication Dates

  • Publication in this collection
    18 Mar 2022
  • Date of issue
    2022

History

  • Received
    13 Oct 2021
  • Accepted
    27 Dec 2021
Sociedade Brasileira de Ciência e Tecnologia de Alimentos Av. Brasil, 2880, Caixa Postal 271, 13001-970 Campinas SP - Brazil, Tel.: +55 19 3241.5793, Tel./Fax.: +55 19 3241.0527 - Campinas - SP - Brazil
E-mail: revista@sbcta.org.br