Cross Validation Voting for Improving CNN Classification in Grocery Products

Domingo, Jaime Duque; Medina, Roberto; Rodrigo, Luis Miguel González

doi:10.1109/access.2022.3152224

Cited by 15 publications

(7 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The model structure can be referred to Figure 14 and Table 15. To further improve the classification accuracy on the CIFAR-10 dataset, we employed cross-validation ensemble for further enhancement, based on the experimental results and experience of J. D. Domingo et al 35 . We adopted a five-fold cross-validation ensemble, and the model structure is shown in Figure 15.…”

Section: Experiments Descriptionmentioning

confidence: 99%

A improved pooling method for convolutional neural networks

Zhao,

Zhang

2024

Sci Rep

View full text Add to dashboard Cite

The pooling layer in convolutional neural networks plays a crucial role in reducing spatial dimensions, and improving computational efficiency. However, standard pooling operations such as max pooling or average pooling are not suitable for all applications and data types. Therefore, developing custom pooling layers that can adaptively learn and extract relevant features from specific datasets is of great significance. In this paper, we propose a novel approach to design and implement customizable pooling layers to enhance feature extraction capabilities in CNNs. The proposed T-Max-Avg pooling layer incorporates a threshold parameter T, which selects the K highest interacting pixels as specified, allowing it to control whether the output features of the input data are based on the maximum values or weighted averages. By learning the optimal pooling strategy during training, our custom pooling layer can effectively capture and represent discriminative information in the input data, thereby improving classification performance. Experimental results show that the proposed T-Max-Avg pooling layer achieves good performance on three different datasets. When compared to LeNet-5 model with average pooling, max pooling, and Avg-TopK methods, the T-Max-Avg pooling method achieves the highest accuracy on CIFAR-10, CIFAR-100, and MNIST datasets.

show abstract

Section: Experiments Descriptionmentioning

confidence: 99%

A improved pooling method for convolutional neural networks

Zhao,

Zhang

2024

Sci Rep

View full text Add to dashboard Cite

show abstract

“…Then, the part presenting the smaller number of windows is kept, while windows of the other part are randomly discarded until the sizes of both groups (DER and non-DER windows) are matched. This creates an even input data (50% DER -50% non-DER), which contributes to training the machine learning algorithms for a more general scenario and reduces the likelihood of overfitting [34].…”

Section: Data Processingmentioning

confidence: 99%

“…During training, 80% of the windows created from the input matrix X and its correspondent element in the output vector y are used to create a model f (x). The model f (x) is then validated using 20% of unseen data from the input matrix and the predictions are compared with the expected values from the vector y at each timestep t. To provide a comprehensive evaluation of the NILM method proposed, cross-validation with a 5 k-fold is used to reduce possibility of overfitting [34].…”

Section: F Nilm Model Assessmentmentioning

confidence: 99%

Distributed Energy Resources Electric Profile Identification in Low Voltage Networks Using Supervised Machine Learning Techniques

et al. 2023

View full text Add to dashboard Cite

Increasing integration of distributed energy resources (DER) in the electrical network has led distribution network operators to unprecedented challenges. This issue is compounded by the lack of monitoring infrastructure on the low voltage (LV) side of distribution networks at residential and utility sides. Non-intrusive load monitoring (NILM) methods provide an opportunity to add value to conventional electric measurements and to increase the observability of LV networks for the implementation of active management network techniques and intelligent control of DER. This work proposes a novel implementation of NILM methods for the identification of DER electrical signatures from aggregated measurements taken at the LV side of a distribution transformer. The implementation evaluates three machine learning algorithms such as k Nearest Neighbours (kNN), random forest and a multilayer perceptron under 100 scenarios of DER integration. A year of minutely reported values of electric current, voltage, active power, and reactive power are used to train and test the proposed model. The F 1 scores achieved of 73% and 93% for Electrical Vehicles (EV) and rooftop photovoltaic (PV) respectively and processing times below 314 µs on an Intel Core i7-8700 machine. These results confirm the relevance of the NILM method based on low frequency electric measurements from the real-time identification of DER.INDEX TERMS Distributed energy resources (DER), non-intrusive load monitoring (NILM), low voltage networks distribution networks (LVN), random forest, supervised machine learning.The associate editor coordinating the review of this manuscript and approving it for publication was Wentao Fan .

show abstract

“…The cross-validation mechanism extends the training procedure by one more step. [8]. It divides the original sample set into six parts.…”

Section: Preprocess the Datamentioning

confidence: 99%

Diagnosis of hepatitis C virus based on ID3 algorithm

Che

2023

ACE

View full text Add to dashboard Cite

Infection with the hepatitis C virus (HCV) is what causes hepatitis C. China has one of the highest hepatitis C infection rates in the world (13%-15%), and this illness affects more than 170 million people worldwide. As a result, there is an enormous need for the diagnosis of this disease. In recent years, researchers have made significant progress in this field with different machine learning algorithms and have had the ability to make a relatively precise diagnosis. However, within the machine learning algorithms that the researchers used, decision trees, particularly the ID3(Iterative Dichotomiser 3) algorithm, have been disregarded. This paper uses this algorithm in the diagnosis of hepatitis C, aiming to test the accuracy and explore the possibility of introducing this approach into application in medicine. Empirically, the novel method of applying the ID3 algorithm in the diagnosis of hepatitis C can produce relatively accurate results compared with those traditional approaches, demonstrating that this method can be used in practice and is a powerful approach for diagnosing people who have the hepatitis C virus.

show abstract

Cross Validation Voting for Improving CNN Classification in Grocery Products

Cited by 15 publications

References 38 publications

A improved pooling method for convolutional neural networks

A improved pooling method for convolutional neural networks

Distributed Energy Resources Electric Profile Identification in Low Voltage Networks Using Supervised Machine Learning Techniques

Diagnosis of hepatitis C virus based on ID3 algorithm

Contact Info

Product

Resources

About