Training Data For Classification Research Articles

Assessing the risks and opportunities posed by AI-enhanced influence operations on social media

AbstractLarge language models (LLMs) like GPT-4 have the potential to dramatically change the landscape of influence operations. They can generate persuasive, tailored content at scale, making campaigns using falsified content, such as disinformation and fake accounts, easier to produce. Advances in self-hosted open-source models have meant that adversaries can evade content moderation and security checks built into large commercial models such as those commercialised by Anthropic, Google, and OpenAI. New multi-lingual models make it easier than ever for foreign adversaries to pose as local actors. This article examines the heightened threats posed by synthetic media, as well as the potential that these tools hold for creating effective countermeasures. It begins with assessing the challenges posed by a toxic combination of automated bots, human-controlled troll accounts, and more targeted social engineering operations. However, the second part of the article assesses the potential for these same tools to improve detection. Promising countermeasures include running internal generative models to bolster training data for internal classifiers, detecting statistical anomalies, identifying output from common prompts, and building specialised classifiers optimised for specific monitoring needs.

Analysis Sentiment to Religious Based Issues Netizen Comments on President Jokowi 's Instagram using the Naive Bayes Classifier Method


 
 
 
 The presence of Instagram is one of the most popular social media platforms. Make people make a habit of posting photo and video content. In this way, Instagram can become a source of public opinion and sentiment data, which data can be used for social studies. Social studies is still a crucial issue for countries today, including Indonesia, is a matter of religion. Apart from problems regarding religion, public sentiment can be researched from Instagram. The sentiment dataset related to religion is taken from public comments from President Joko Widodo's content regarding Merry Christmas Greetings and Visits to Churches. The analysis system will use sentiment analysis using a method from machine learning, namely the naive Bayes classifier to determine positive & negative sentiments. The comment dataset is 2,600 in which the positive and negative sentiment class datasets are determined manually by experts in the field of linguistics or communication totaling 1,309 to be used as training data. the results of testing the Naïve Bayes classifier training data with a total of 1309 obtained an accuracy prediction of 97.63% and test data where positive & negative sentiment classes had not been determined before hand amounted to 1291 in sentiment analysis obtained positive sentiment 70.56% & negative sentiment 29.44%.
 
 
 

Gray Level Co-Occurrence Matrices and Support Vector Machine for Improved Lung Cancer Detection

A detection system based on digital image processing and machine learning classification was developed to detect normal and cancerous lung conditions. 340 data from LIDC –IDRI were processed through several stages. The first stage is pre-processing using three filter variations and contrast stretching, which reduce noise and increase image contrast. The image segmentation process uses Otsu Thresholding to clarify the ROI of the image. The texture feature extraction with GLCM was applied using 21 feature variations. Data extraction is used as a label value learned by the classification system in the form of SVM. The results of the training data classification are processed with a confusion matrix which shows that the high pass filter has higher accuracy than the other two variations. The proposed method was assessed in terms of accuracy, precision and recall. The model provided an accuracy of 99.67 % training data and 97.50 % testing data.

Classification of Basurek Batik Using Pre-Trained VGG-16 and Support Vector Machine

By introducing Indonesian batik motifs, we know that the island of Sumatra, especially Bengkulu and Jambi provinces, has a distinctive batik called Basurek batik. This research aims to classify the two batik motifs using the Support Vector Machine (SVM) algorithm. First, we extract the image of the batik motif with a pre trained VGG-16 model and then use them as a dataset for the SVM classification process. The classification process itself uses linear, polynomial, and sigmoid kernels. We divided the data 90:10 and used 10-fold cross-validation to analyze each training and testing data classification result. The results of this study are the highest values of accuracy, precision, and recall of 76.4%, 76.5%, and 76.4% produced by the linear kernel for the training data classification. For the testing data classification, both the linear and polynomial kernels generate the best accuracy, precision, and recall values of 87.5%, 90%, and 85.5%. On average, incorporating the training and testing classification results, we found that the linear kernel is the best function for classifying the Basurek batik motif using the collected images from the internet.

Bulldog Breed Classification Using VGG-19 and Ensemble Learning

In image classification, the C4.5, Adaboost, and Gradient Boosting algorithms need another method to extract the image's features in the classification process. This research employs transfer learning with the VGG-19 network for the image's features extraction and transfers the result as a dataset to classify image-based Bulldog breeds. As the classifier to classify the extracted features from the VGG 16 model, we employ three ensemble learning algorithms, namely C4.5, AdaBoost, and Gradient Boost. The training data classification results of the American, English, and French bulldog breeds show that, with a 20-fold cross-validation evaluation, the Gradient Boosting algorithm performs the best, with an accuracy value of 0.958, a precision value of 0.958 and recall value of 0.933. And show the highest accuracy (0.933), precision (0.938), and recall (0.933) in the testing data classification. While in the testing data classification, the Gradient Boosting algorithm scores an accuracy value of 0.933, a precision value of 0.938, and a recall value of 0.933.

Covid-19 Fake News Detection on Twitter Based on Author Credibility Using Information Gain and KNN MethodsCovid-19 Fake News Detection on Twitter Based on Author Credibility Using Information Gain and KNN Methods

Twitter is one of the social media that is used as a tool to share various kinds of information about various kinds of things that are of concern to social media users. One of the information shared is information about COVID-19, which is known that the COVID-19 pandemic is currently spreading throughout the world at a very alarming rate. COVID-19 is an infectious disease caused by SARS-COV-2. The World Health Organization (WHO) claims that the spread of COVID-19 is supported by the spread of false/fake news. So to find out the truth of the news, a COVID-19 fake news detector is needed so that users don't fall for the hoaxes circulating. This study aims to classify COVID-19 news on Twitter based on author credibility. Credibility in question is a person's perception of the validity of information and is a multidimensional concept that is used as a means of receiving information to assess the source of communication. The method used in this research is Information Gain and KNN. KNN (K-Nearest Neighbor) is a supervised learning algorithm that works by classifying a set of data based on classified training data. Information Gain is used to ranking the most influential attributes, and KNN is used to classify data based on learning data taken from the nearest neighbors. The research consists of 6 main stages, namely data collection (crawling data), data preprocessing, feature extraction, feature selection, data split into training data and testing data, KNN stage, and data evaluation stage. The research carried out succeeded in obtaining an accuracy value of 91%, a correlation value between credibility and hoax of 0.115, and a p-value <0.005.

인공지능 학습모델과 음향데이터를 이용한 군사장비 식별

Acoustic data is widely used for the development of voice recognition based services, speaker identification system, or environmental sound classification. However, in the defense sector, research on the use of acoustic data and AI learning models has not yet been actively conducted. In addition, collection methods or quality standards that must be considered in order to use acoustic data as training data have not been clearly established. Therefore, this paper attempts to classify six weapons for the first time using four AI learning models and previously collected acoustic data to derive considerations in the process of collecting acoustic data. The learning data used in the experiment is acoustic data related to Army equipment collected through a government-supported project in the second half of 2021. Experimental results show that the classification performance of CNN and LightGBM models is generally high and the best performance is achieved when the data has a medium length. By uniformly re-configuring the number of data per class using SMOTE, the performance can be further improved. Based on these experimental results, we also propose what should be considered when constructing AI training data for classification of military equipment based on acoustic data.

LiDAR-Based Hand Contralateral Controlled Functional Electrical Stimulation System.

Contralateral controlled functional electrical stimulation (CCFES) can induce simultaneous movements in patients' bilateral hands. It has been clinically proven to be effective in improving hand motor control and dexterity. sEMG and bending sensor-based data gloves for detecting patients' motor intent have been developed with limitations. sEMG sensor signals are unstable and susceptible to noise. Data gloves composed of bending sensors require complicated calibration and tend to have data drift. In this paper, a LiDAR-based system for hand CCFES is proposed. The method utilized LiDAR to detect the patient's motion intention without contact in CCFES systems. It has been clinically proven that LiDARs can effectively distinguish the different motion amplitudes of hand gestures as quantitative evaluation sensors of functional electrical stimulation (FES). Training data for classifiers were collected from 9 healthy individuals and 15 stroke patients performing 4 gestures, including hand opening, fist clenching, wrist extension, and wrist flexion. The support vector machine (SVM), linear discriminant analysis (LDA), and k-nearest neighbor (kNN) were verified for their classification performance in offline hand gesture recognition tests. Experiments were also conducted on 6 stroke volunteers to evaluate gestures triggered by FES. The SVM classifier showed excellent classification performance for four hand gestures, with an average F1-score of 0.97 ± 0.05 in offline tests. As for online gesture recognition, an average F1-score of 0.92 ± 0.09 was obtained. In the evaluation experiments, between data from 50% and 100% movement amplitude, paired t-tests showed significant differences. The experimental results indicated that the proposed system showed promise for hand rehabilitation.

An Intelligent Admission Control Scheme for Dynamic Slice Handover Policy in 5G Network Slicing

5G use cases, for example enhanced mobile broadband (eMBB), massive machine-type communications (mMTC), and an ultra-reliable low latency communication (URLLC), need a network architecture capable of sustaining stringent latency and bandwidth requirements; thus, it should be extremely flexible and dynamic. Slicing enables service providers to develop various network slice architectures. As users travel from one coverage region to another area, the call must be routed to a slice that meets the same or different expectations. This research aims to develop and evaluate an algorithm to make handover decisions appearing in 5G sliced networks. Rules of thumb which indicates the accuracy regarding the training data classification schemes within machine learning should be considered for validation and selection of the appropriate machine learning strategies. Therefore, this study discusses the network model’s design and implementation of self-optimization Fuzzy Q-learning of the decision-making algorithm for slice handover. The algorithm’s performance is assessed by means of connection-level metrics considering the Quality of Service (QoS), specifically the probability of the new call to be blocked and the probability of a handoff call being dropped. Hence, within the network model, the call admission control (AC) method is modeled by leveraging supervised learning algorithm as prior knowledge of additional capacity. Moreover, to mitigate high complexity, the integration of fuzzy logic as well as Fuzzy Q-Learning is used to discretize state and the corresponding action spaces. The results generated from our proposal surpass the traditional methods without the use of supervised learning and fuzzy-Q learning.

Detection of passageways in natural foliage using biomimetic sonar

The ability of certain bat species to navigate in dense vegetation based on trains of short biosonar echoes could provide for an alternative parsimonious approach to obtaining the sensory information that is needed to achieve autonomy in complex natural environments. Although bat biosonar has much lower data rates and spatial (angular) resolution than commonly used human-made sensing systems such as LiDAR or stereo cameras, bat species that live in dense habitats have the ability to reliably detect narrow passageways in foliage. To study the sensory information that the animals may have available to accomplish this, we have used a biomimetic sonar system that was combined with a camera to record echoes and synchronized images from 10 different field sites that featured narrow passageways in foliage. The synchronized camera and sonar data allowed us to create a large data set (130 000 samples) of labeled echoes using a teacher–student approach that used class labels derived from the images to provide training data for echo-based classifiers. The performance achieved in detecting passageways based on the field data closely matched previous results obtained for gaps in an artificial foliage setup in the laboratory. With a deep feature extraction neural network (VGG16) a foliage-versus-passageway classification accuracy of 96.64% was obtained. A transparent artificial intelligence approach (class-activation mapping) indicated that the classifier network relied heavily on the initial rising flank of the echoes. This finding could be exploited with a neuromorphic echo representation that consisted of times where the echo envelope crossed a certain amplitude threshold in a given frequency channel. Whereas a single amplitude threshold was sufficient for this in the previous laboratory study, multiple thresholds were needed to achieve an accuracy of 92.23%. These findings indicate that despite many sources of variability that shape clutter echoes from natural environments, these signals contain sufficient sensory information to enable the detection of passageways in foliage.

Software Package for Efficient Creation of Training Data for Machine Learning Classifiers from Micrographs

An abstract is not available for this content so a preview has been provided. As you have access to this content, a full PDF is available via the ‘Save PDF’ action button.

Application of Machine Learning to Determine the Factors Affecting Deterioration in Patients with Chronic Kidney Disease

Hospital databases generally contain large amounts of data and various, but it has not been used optimally. It needs a technique that can utilize mountains of data into strategically valuable information. This paper will investigate ways to use hospital data to help determine the factors that influence the deterioration in patients with chronic kidney disease. The criteria for the selected patients were patients with a diagnosis of chronic kidney disease and chemotherapy treatment at least once. Three hundred seventy-six patients met these criteria. Subsequently, observation the patient's treatment course for three years. Ninety patients died in the hospital during that period. All the results of patients' blood tests were collected for further analysis. In forming the classification model, there are three stages carried out. The first stage deals with diverse, incomplete, and inconsistent data. Then through the process of changing continuous data into categorical data, each variable is classified into several categories. The next stage is to create a predictive model to determine the factors that influence the deterioration in patients with kidney failure using the Random Forest, Logistic Regression, and Decision Tree algorithms. Information of the classification model, 12 variables were selected, namely age, sex, and the results of clinical pathology laboratory examinations-Ureum, Thrombocyte, Natrium, Creatinine, Chloride, Kalium, Hemoglobin, Hematocrit, and Leukocytes. The three algorithms can classify training data with an accuracy of 98% (Random Forest), 83% (Logistic Regression), 98% (ID3).

CLASSIFICATION SUPPORT VECTOR MACHINE IN BREAST CANCER PATIENTS

Support vector machine is one of the supervised learning methods in machine learning that is used in classification. The purpose of this study is to measure the accuracy of classification by using 3 hyperplane functions in SVM, namely linear, sigmoid, polynomial, and radial basis function (RBF). Based on the simulation results of training data and testing data on female breast cancer patients, SVM with hyperplane RBF has better accuracy than hyperplane polynomial, linear and sigmoid. The RBF results for the training and testing data were 89.1% and 73.2%, respectively. Based on the results of the classification of training data for female breast cancer patients, 88.07% had no recurrence and 93.33% had recurrence events. Meanwhile, based on the results of the classification of testing data, female patients did not recurrence events and recurrence events was 72.55% and 80.00%, respectively. So from this article, it can be concluded that SVM with hyperplane RBF is one of the best methods in the application of the method of classifying female breast cancer patients.

Comparison of the Accuracy of Sentiment Analysis on the Twitter of the DKI Jakarta Provincial Government during the COVID-19 Vaccine Time

Currently, the Government is intensively utilizing social media, one of which is Twitter as a place of interaction with the community. The results of these interactions can be used as feedback to determine whether public opinion on public policies is positive or negative. Tweets from users can be a supporting parameter for the government in evaluating future policies and decision making by applying the sentiment analysis method. This study aims to determine positive or negative sentiments on user tweets against the official twitter account of the DKI Jakarta Provincial Government during the COVID19 vaccine period. The data obtained are 1658 lines from March 30 to April 5, 2021 with queries on tweets containing words or mentioning the username @dkijakarta, which will be grouped by sentiment class, namely negative and positive using the TF-IDF Vectorizer for word weighting and classification using several methods, namely, nave Bayes with accuracy values. 82.50% with class recall on positive sentiment 88% and negative 77% and in class precision showing positive at 79.28% and negative at 86.52% in the rapid miner application then k-NN with an accuracy value of 81.50% with class recall on positive sentiment 85% and negative 78% and class precision shows positive at 79.44% and negative at 83.87% in the rapid miner application. And the accuracy value of the best method in this training data classification comparison is nave Bayes, the results the end of testing the sample dataset using the nave Bayes method with 84.80% accuracy with class recall at 85.01% positive sentiment and 84.59% negative sentiment and at c lass precision shows positive at 85.21% and negative at 84.38% in rapid mining applications.

Closing the loop: automatically identifying abnormal imaging results in scanned documents.

Scanned documents (SDs), while common in electronic health records and potentially rich in clinically relevant information, rarely fit well with clinician workflow. Here, we identify scanned imaging reports requiring follow-up with high recall and practically useful precision. We focused on identifying imaging findings for 3 common causes of malpractice claims: (1) potentially malignant breast (mammography) and (2) lung (chest computed tomography [CT]) lesions and (3) long-bone fracture (X-ray) reports. We train our ClinicalBERT-based pipeline on existing typed/dictated reports classified manually or using ICD-10 codes, evaluate using a test set of manually classified SDs, and compare against string-matching (baseline approach). A total of 393 mammograms, 305 chest CT, and 683 bone X-ray reports were manually reviewed. The string-matching approach had an F1 of 0.667. For mammograms, chest CTs, and bone X-rays, respectively: models trained on manually classified training data and optimized for F1 reached an F1 of 0.900, 0.905, and 0.817, while separate models optimized for recall achieved a recall of 1.000 with precisions of 0.727, 0.518, and 0.275. Models trained on ICD-10-labelled data and optimized for F1 achieved F1 scores of 0.647, 0.830, and 0.643, while those optimized for recall achieved a recall of 1.0 with precisions of 0.407, 0.683, and 0.358. Our pipeline can identify abnormal reports with potentially useful performance and so decrease the manual effort required to screen for abnormal findings that require follow-up. It is possible to automatically identify clinically significant abnormalities in SDs with high recall and practically useful precision in a generalizable and minimally laborious way.

Face mask detection services of Covid19 monitoring system to maintain a safe environment using deep learning method

COVID19 has become one of the pandemic diseases that hit the world, including in Indonesia. Efforts to prevent the spread of this virus have been carried out, including by implementing health protocols, provide information services for the spread of the virus and emergency response, detection services for people suspected and infected with the virus and programs for preventing the spread of the virus and vaccination for all elements of society in Indonesia. This research focuses on the development of face mask detection system to detect the use of mask by people using deep learning method that focuses more on object detection. Face masked detection systems can be used to assist in monitoring compliance with the use of masks in public areas that have the potential to cause crowds so as to create a safe environment. The use of the Deep Learning method through the MobileNets V2 architecture was quite effective and gives an accuracy result in testing the classification training data of 0.99 at 10 epochs and has also been implemented in real-time direct testing.

Analisis Data mining dengan Metode C.45 pada Klasifikasi Kenaikan Rata-Rata Volume Perikanan Tangkap

Increasing the amount of demand for natural resource needs is increasing. One of them is natural resources in the sea and coast. The current condition of capture fisheries in Indonesia is not yet optimal. This is indicated by the increase in the volume of capture fisheries production which is very slow. The purpose of this study is to make data classification for the prediction of the average volume increase in capture fisheries with data mining techniques. Data mining techniques are applied to determine the data patterns of the capture fisheries dataset, so the results of the classification can be applied to evaluate the factors that affect the volume of capture fisheries. The classification algorithm used is C45. The results of the classification were tested with rapidminer in classifying data. The level of performance is indicated by the accuracy value. The accuracy value is obtained by testing the results of the classification of training data and testing data. Comparison of accuracy values between the algorithms used can be seen the best algorithm in making the classification of capture fisheries data.

Classification of Myocardial 18F-FDG PET Uptake Patterns Using Deep Learning.

To perform automated myocardial segmentation and uptake classification from whole-body fluorine 18 fluorodeoxyglucose (FDG) PET. In this retrospective study, consecutive patients who underwent FDG PET imaging for oncologic indications were included (July-August 2018). The left ventricle (LV) on whole-body FDG PET images was manually segmented and classified as showing no myocardial uptake, diffuse uptake, or partial uptake. A total of 609 patients (mean age, 64 years ± 14 [standard deviation]; 309 women) were included and split between training (60%, 365 patients), validation (20%, 122 patients), and testing (20%, 122 patients) datasets. Two sequential neural networks were developed to automatically segment the LV and classify the myocardial uptake pattern using segmentation and classification training data provided by human experts. Linear regression was performed to correlate findings from human experts and deep learning. Classification performance was evaluated using receiver operating characteristic (ROC) analysis. There was moderate agreement of uptake pattern between experts and deep learning (as a fraction of correctly categorized images) with 78% (36 of 46) for no uptake, 71% (34 of 48) for diffuse uptake, and 71% (20 of 28) for partial uptake. There was no bias in LV volume for partial or diffuse uptake categories (P = .56); however, deep learning underestimated LV volumes in the no uptake category. There was good correlation for LV volume (R 2 = 0.35, b = .71). ROC analysis showed the area under the curve for classifying no uptake and diffuse uptake was high (> 0.90) but lower for partial uptake (0.77). The feasibility of a myocardial uptake index (MUI) for quantifying the degree of myocardial activity patterns was shown, and there was excellent visual agreement between MUI and uptake patterns. Deep learning was able to segment and classify myocardial uptake patterns on FDG PET images.Keywords: PET, Heart, Computer Aided Diagnosis, Computer Application-Detection/DiagnosisSupplemental material is available for this article.©RSNA, 2021.

Long-Term BCI Training of a Tetraplegic User: Adaptive Riemannian Classifiers and User Training.

While often presented as promising assistive technologies for motor-impaired users, electroencephalography (EEG)-based Brain-Computer Interfaces (BCIs) remain barely used outside laboratories due to low reliability in real-life conditions. There is thus a need to design long-term reliable BCIs that can be used outside-of-the-lab by end-users, e.g., severely motor-impaired ones. Therefore, we propose and evaluate the design of a multi-class Mental Task (MT)-based BCI for longitudinal training (20 sessions over 3 months) of a tetraplegic user for the CYBATHLON BCI series 2019. In this BCI championship, tetraplegic pilots are mentally driving a virtual car in a racing video game. We aimed at combining a progressive user MT-BCI training with a newly designed machine learning pipeline based on adaptive Riemannian classifiers shown to be promising for real-life applications. We followed a two step training process: the first 11 sessions served to train the user to control a 2-class MT-BCI by performing either two cognitive tasks (REST and MENTAL SUBTRACTION) or two motor-imagery tasks (LEFT-HAND and RIGHT-HAND). The second training step (9 remaining sessions) applied an adaptive, session-independent Riemannian classifier that combined all 4 MT classes used before. Moreover, as our Riemannian classifier was incrementally updated in an unsupervised way it would capture both within and between-session non-stationarity. Experimental evidences confirm the effectiveness of this approach. Namely, the classification accuracy improved by about 30% at the end of the training compared to initial sessions. We also studied the neural correlates of this performance improvement. Using a newly proposed BCI user learning metric, we could show our user learned to improve his BCI control by producing EEG signals matching increasingly more the BCI classifier training data distribution, rather than by improving his EEG class discrimination. However, the resulting improvement was effective only on synchronous (cue-based) BCI and it did not translate into improved CYBATHLON BCI game performances. For the sake of overcoming this in the future, we unveil possible reasons for these limited gaming performances and identify a number of promising future research directions. Importantly, we also report on the evolution of the user's neurophysiological patterns and user experience throughout the BCI training and competition.

Short-term wind power forecasting model based on multi-feature extraction and CNN-LSTM

Improving the accuracy of short-term wind power forecasting is critical to wind power consumption. This paper establishes a short-term wind power prediction model based on the multi-feature extraction and deep learning network CNN-LSTM. Muti-features are extracted from original data to improve the accuracy of training. In addition, clustering algorithm is used to classify training data and train the models corresponding to those classes. CNN-LSTM prediction models are established for each cluster and compared with ARIMA, RNN, CNN and LSTM models.