This text is a survey on cross-validation We define all classical cross-validation procedures, and we study their properties for two different goals: estimating the risk of a given estimator, and selecting the best estimator among a given family For the risk estimation problem, we compute the bias (which can also be corrected) and the variance of cross-validation methods For estimator selection, we first provide a first-order analysis (based on expectations) Then, we explain how to take into account second-order terms (from variance computations, and by taking into account the usefulness of overpenalization) This allows, in the end, to provide some guidelines for choosing the best cross-validation method for a given learning problem

Cross-validation

Background: The COVID-19 pandemic has made people spend more time on online meetings more than ever. The prolonged time looking at the monitor may cause fatigue, which can subsequently impact the mental and physical health. A fatigue detection system is needed to monitor the Internet users well-being. Previous research related to the fatigue detection system used a fuzzy system, but the accuracy was below 85%. In this research, machine learning is used to improve accuracy. Objective: This research examines the combination of the FaceNet algorithm with either k-nearest neighbor (K-NN) or multiclass support vector machine (SVM) to improve the accuracy. Methods: In this study, we used the UTA-RLDD dataset. The features used for fatigue detection come from the face, so the dataset is segmented using the Haar Cascades method, which is then resized. The feature extraction process uses FaceNet's pre-trained algorithm. The extracted features are classified into three classes—focused, unfocused, and fatigue—using the K-NN or multiclass SVM method. Results: The combination between the FaceNet algorithm and K-NN, with a value of  resulted in a better accuracy than the FaceNet algorithm with multiclass SVM with the polynomial kernel (at 94.68% and 89.87% respectively). The processing speed of both combinations of methods has allowed for real-time data processing. Conclusion: This research provides an overview of methods for early fatigue detection while working at the computer so that we can limit staring at the computer screen too long and switch places to maintain the health of our eyes.

/pdf/fatigue-detection-on-face-image-using-facenet-algorithm-and-2gtznk27b3.pdf

Fatigue Detection on Face Image Using FaceNet Algorithm and K-Nearest Neighbor Classifier

A solution has been developed due to problem of face recognition and actions when we are trying to process images of human faces in real time on the server. Approaches to solving the problem of face recognition and actions on images are considered. The solution of the problem of action’s recognition in real time is considered. An action recognition system was used for proctoring.A hybrid architecture was developed for face and action recognition. The FaceNet algorithm was used to implement the system. We used the faces of a person that was filmed to collect the dataset. The results obtained show a high accuracy of the face recognition percentage. Face and motion recognition is implemented in Windows Server 2016. A special web application was developed to test these algorithms.To reduce processing time and recognition algorithms were used like bandwidth, resize images. In addition, the entire process of video processing for proctoring was launched in the graphics server. With the web app, you can see how well the face recognition algorithm and the result of the action recognition algorithm are.

Hybrid Architecture of Face and Action Recognition Systems for Proctoring on a Graphic Processor

As well known, there has always been a strong connection between the attendance of a student being linked to their performance which indeed refers to their success ultimately, and seminars play an important role when it comes to assisting the students to meet industries’ expectations. And the traditional way of collecting the attendance before seminars is time-consuming, thus making less time for the presenter, and this can be solved using face recognition technology. Whenever there is a seminar to be held, all eligible members should register themselves in the proposed activity system with all their credentials along with their facial image. As the seminar starts, the admin captures the images of the members who are attending the seminar, and these images are used for attendance by using face recognition technology. Here, the Histogram of Oriented Gradients is used for face detection and the K- Nearest Neighbors algorithm is used for face recognition which has shown the highest accuracy. The accuracy of the algorithm in this project has shown to be 97%. The proposed system along with providing the list of all attended members also provides the attended members with a feedback slot where the members can provide feedback in the form of text. Using Natural Language Processing, Sentiment Analysis when performed on the feedback gives a clear picture of the opinions of the members, represented in the form of a histogram showing the number of members who opted for positive or negative or neutral feedback.

A Face Recognition and Sentiment Analysis Activity System using Machine Learning Algorithm

The human face is an example of the unique biometric data that each person has. The human face contains a lot of information, including face shape, skin color, eye shape, nose shape, mouth shape, and some additional attributes such as beard, mustache, hair, and eyebrows. This biometric information can be used to obtain further information regarding the human identity itself. In this study, sample data of human faces aged 20 to 30 years were used. The purpose of this study is to classify gender in human facial images using masks, while the algorithm used uses supervised learning. The research process carried out was collecting human face sample data, conducting training stages which included face resizing, storing data in arrays, feature extraction using faceNet, training data and gender models. Perform testing stages which include video data acquisition, video extraction into frames, video frames, face detection, facial image resixing, storing data in arrays, feature extraction using faceNet and prediction of gender determination. The benefits of this research can provide information in determining the gender of human faces with masks, apply the application of supervised learning algorithms to processing human face images and can be applied to face detection on highways, public areas and shopping centers. Based on testing using the K-NN method, the train accuracy value is above 87% and the test accuracy is above 96%. Testing using the SVM method obtained train accuracy above 99% and test accuracy above 98%. Tests using the random forest method obtained 100% train accuracy and test accuracy above 88%.

/pdf/klasifikasi-jenis-kelamin-wajah-bermasker-menggunakan-25k58wso.pdf

Klasifikasi Jenis Kelamin Wajah Bermasker Menggunakan Algoritma Supervised Learning

This study discusses the appropriate method to be applied in a presence system using faces by comparing two deep learning architectural models, they are FaceNet and Openface. FaceNet is a model developed by Google researchers that has the highest accuracy in face recognition. While Openface is a development from FaceNet that is trained with smaller datasets but has an accuracy that is almost equal to FaceNet. This will start by taking the employee's face into an image dataset. From the dataset, the face preprocessing will be performed by detecting, cropping, and resizing the face. Then extracting facial features into 128 dimensions using the FaceNet and Openface. With the Support Vector Machine (SVM), the classification of facial features will be carried out to obtain accuracy. To validate the model, 5 fold cross-validations are used. FaceNet accuracy results that obtained are higher with perfect accuracy that is 100%, while Openface only 93.33% accuracy. The implementation using the model with the highest accuracy (FaceNet) has the same results as the model testing that is 100% using the introduction threshold probability of 0.25.

Ferry Cahyono

Papers

Face Recognition System using Facenet Algorithm for Employee Presence