An Improved Precision Using an Integrated Neural Network Model and Support Vector Machine Algorithms to Find the Instagram Users Profile Classification

Published: Dec. 14, 2023

This work proposes a Novel Integrated Neural Network (NINN) for Fake profile detection on Instagram and evaluates its precision against Support Vector Machine (SVM) algorithm. The goal is to enhance the accuracy of identifying fake profiles social media platform using NINN approach. It expected that will provides reliable predictions, with greater accurate results than existing models considered in this article classifying reducing security threats Instagram. research utilized two sample groups, each containing 20 samples. We used clinic calc tool perform calculations set significance level at 0.05. implies we are ready tolerate 5% probability rejecting null hypothesis when it true. also specified last beta rate 0.2 95% confidence interval (CI) effect size estimation. trained dataset cost function, their performance measured by output. model has three components: an input component, hidden component implements algorithm, output shows After conducting thorough analysis, Networks Algorithm was found have 90.91%, while had 88.19%. A statistical analysis performed compare algorithms. Independent samples T-Test, which tests means populations equal. yielded p-value p=0.000 (p<0.05), indicates can be rejected high confidence. Therefore, difference between algorithms statistically significant. In Automated classification Instagram, algorithm higher prediction percentage (90.91%) (88.19%). more precise method implemented applying along SVM allows users distinguish accurately. study overcome drawbacks approaches developed better approach identify fraudulent accounts

Language: Английский

---

Recognizing Distraction for Assistive Driving by Tracking Body Parts Using Novel Convolutional Neural Network with LGM Classifier Over Random Forest with Improved Accuracy

Published: Dec. 14, 2023

In the context of identifying driver distractions, goal this study is to investigate degree accuracy exhibited by most recent iterations deep learning algorithms. Both Convolutional Neural Networks Utilising LGM Classifier (CNNLGM) and Random Forest (RF) are compared head-to-head in research presented here. The investigation required a total 118 samples, which were then divided equally between two categories consisting 58 specimens each. Group 1 utilized CNNLGM Classifier, contrast 2's utilization RF technique. code was implemented using software from Google Colab, same program also used import dataset. A pre-test power 80% an alpha value 0.05 taken into consideration while determining appropriate size sample for experiment. This accomplished with assistance online tool statistical analysis. Previous provided necessary information use. findings simulation showed that Novel achieved 96%, whereas algorithm could only achieve 82%. There substantial disparity levels approaches, as measured significance 0.001 (p<0.05). To summarise, when utilizing data supplied, performed noticeably better than it came paying attention.

Language: Английский

---

Enhancing Accuracy of Bitcoin Cost Prediction using Ridge Linear Classification over Lasso Regression

Published: Dec. 14, 2023

Enhancing accuracy of Bitcoin cost prediction using ridge linear classification over lasso-regression. For confirmation purposes, use the financial terms 'sustain' and 'conflict' repeatedly to compare forecast. A simple method called Lasso-regression is used forecast price bit-coin. Evaluation was conducted assess usefulness each model's presentation for task, results were analyzed. The aim trying so many dissimilar models analyze variations in their essential hypotheses. In order reflect on contribution variables, we finest parameters, which a trusted practice. model that derived from T-test has value p 0.002, (p<0.05). This indicates there statistically momentous difference between two algorithms. sample size calculation performed an 80 percent G-power pretest, 0.05% threshold, 95% confidence interval (CI). predicting Bitcoin, improvement 93.06% when ridge-linear regression with enhanced precision lasso regression, as indicated by this research study. Conclusion: Using mapped precision.

Language: Английский

---

Enhancing the Prediction of Communication Using Support Vector Machine Over Decision Tree for Underwater Wireless Networks

Published: Dec. 14, 2023

Effort of this research is to enhance our ability forecast how underwater wireless sensor networks would behave for communication. Materials and make accurate projections on the functionality (UWSN), it necessary put Support Vector Machine (SVM) Decision Tree algorithms through their paces by using a wide range training testing strategies. A sample size 20 both groups calculated Gpower value 85% (g power setting parameters: =0.05 power=0.85). The accuracy based result suggests that SVM, has 92.509% than Tree, having 83.1156%, with statistically significant 0.001 (p<0.05) which concludes higher Accuracy Loss. When compared decision tree, support vector machine superior.

Language: Английский

---

Fake News Classification in Twitter Data Using Innovative K Nearest Neighbor Comparing Logistics Regression

Published: Dec. 14, 2023

Improve the accuracy of identifying fake news on Twitter by employing Advanced K Nearest Neighbor Algorithm then evaluating results against those obtained through Logistic Regression Algorithm. Materials and Methods: The research is divided into two groups, each with a sample size 42 individuals. first group, comprising 21 participants, will apply Algorithm, while second also consisting use technique. study has been designed G power 80% parameters α=0.05 beta=0.2. Result: Innovative nearest neighbor 81.55 % identifies objects increases measured over 79.0 implication value 0.001 (p < 0.05). Conclusion: In terms accuracy, outperforms

Language: Английский

---

Determining Accuracy in Classifying Handwritten Digit Recognition using Support Vector Machine with Freeman Chain Code over K-nearest neighbor

Published: Dec. 14, 2023

The objective of this research project was to distinguish handwritten digits using the Support Vector (SVM) with Freeman chain code, and subsequently, compare prediction accuracy that K-Nearest Neighbor (KNN) algorithm. In pursuit goal, SVM pitted against KNN for task digit detection. Each two experimental groups consisted 20 samples, a pretest power analysis executed an 80% confidence level. This Study revealed Machine achieved 93.48%, whereas algorithm exhibited higher 98.05%. Statistical analysis, performed through Independent Sample T-tests, demonstrated difference in between algorithms is 0.001 (p<0.05). Consequently, it can be concluded surpasses realm recognition.

Language: Английский

---