Enhancing Material Property Predictions through Optimized KNN Imputation and Deep Neural Network Modeling

IgMin Research, Journal Year: 2024, Volume and Issue: 2(6), P. 425 - 431

Published: June 13, 2024

In materials science, the integrity and completeness of datasets are critical for robust predictive modeling. Unfortunately, material frequently contain missing values due to factors such as measurement errors, data non-availability, or experimental limitations, which can significantly undermine accuracy property predictions. To tackle this challenge, we introduce an optimized K-Nearest Neighbors (KNN) imputation method, augmented with Deep Neural Network (DNN) modeling, enhance predicting properties. Our study compares performance our Enhanced KNN method against traditional techniques—mean Multiple Imputation by Chained Equations (MICE). The results indicate that achieves a superior R² score 0.973, represents significant improvement 0.227 over Mean imputation, 0.141 MICE, 0.044 imputation. This enhancement not only boosts but also preserves statistical characteristics essential reliable predictions in science.

Language: Английский

---

Prediction of Cervical Cancer With Application of Machine Learning Models

Advances in healthcare information systems and administration book series, Journal Year: 2024, Volume and Issue: unknown, P. 211 - 221

Published: June 30, 2024

Cancer accounts for a large number of fatalities each year. Cervical cancer is type that starts in the cervix. . very curable and linked to long survival high quality life when detected early. can be prevented by screening tests, such Pap smear test used identify precancerous stages. Nonetheless, there are few disheartening drawbacks includes its poor slide preparation rate human error. Consequently, computer-aided diagnosis system presented as fix issue. Artificial intelligence has been employed over healthcare industry recently, greatly facilitating accurate widespread use medical networks. plays crucial role early cervical cancer. classified normal or abnormal using deep learning machine techniques. This chapter proposes prediction associating classifiers publicly available data set based on risk factors.

Language: Английский

---

Intelligent Detection and Identification of Attacks in IoT Networks Based on the Combination of DNN and LSTM Methods with a Set of Classifiers

Open Journal of Applied Sciences, Journal Year: 2024, Volume and Issue: 14(08), P. 2296 - 2319

Published: Jan. 1, 2024

Language: Английский

---

Enhancing random forest predictive performance for foot and mouth disease outbreaks in Uganda: a calibrated uncertainty prediction approach for varying distributions

Frontiers in Artificial Intelligence, Journal Year: 2024, Volume and Issue: 7

Published: Nov. 1, 2024

Foot-and-mouth disease poses a significant threat to both domestic and wild cloven-hoofed animals, leading severe economic losses jeopardizing food security. While machine learning models have become essential for predicting foot-and-mouth outbreaks, their effectiveness is often compromised by distribution shifts between training target datasets, especially in non-stationary environments. Despite the critical impact of these shifts, implications outbreak prediction been largely overlooked. This study introduces Calibrated Uncertainty Prediction approach, designed enhance performance Random Forest outbreaks across varying distributions. The approach effectively addresses calibrating uncertain instances pseudo-label annotation, allowing active learner generalize more domain. By utilizing probabilistic calibration model, pseudo-annotates most informative instances, refining iteratively minimizing need human annotation outperforming existing methods known mitigate shifts. reduces costs, saves time, lessens dependence on domain experts while achieving outstanding predictive performance. results demonstrate that significantly enhances environments, an accuracy 98.5%, Area Under Curve 0.842, recall 0.743, precision 0.855, F1 score 0.791. These findings underscore Prediction's ability overcome vulnerabilities ML models, offering robust solution contributing broader field modeling infectious management.

Language: Английский

---

DR-Z2AN: dual-recurrent neural network with a tri-channel attention mechanism for financial management prediction

Complex & Intelligent Systems, Journal Year: 2024, Volume and Issue: 11(1)

Published: Nov. 26, 2024

Financial management prediction, often known as financial forecasting, is the act of estimating future outcomes using past data and present trends. It an essential component analysis planning that aids businesses in making well-informed decisions preparing for potential events. In healthcare domain, prediction a crucial task helps patients track predict expenses required their medical services. The established methods have some flaws, such requirement labeled data, quality, time complexity, under fitting problems, longer execution times. Therefore, order to resolve these limitations; deep learning-based model developed this study efficient prediction. Specifically, research proposes dual-recurrent neural network with tri-channel attention mechanism (DR-Z2AN) accurate proposed DR-Z2AN combines dual-RNN multi-head attention, which enhances robustness interpretability systems. learns complex relationships between develops generalization capability tasks. combined efficiently processes sequence improves model's capacity extract meaningful characteristics from input. integration incentive learning approach improve parameters get better results minimum error. experimental demonstrate attains minimal error terms MAE, MAPE, MSE, RMSE 1.46, 3.83, 4.32, 2.08, respectively; thus, gives than other traditional methods. Overall, offers predictions reduced computational improved interpretability.

Language: Английский

---