Machine Learning Techniques for Intelligent Computer Applications: A Unified Performance– Efficiency Framework for Scalable Intelligent Systems

Intelligent computer applications increasingly rely on machine learning techniques to enable autonomous decision-making, adaptive learning, and predictive intelligence across domains such as healthcare analytics, cybersecurity, autonomous systems, and smart infrastructure. Despite rapid progress, existing studies predominantly focus on isolated algorithms or domain-specific applications, limiting their comparative and practical relevance. Thismanuscript presents a comprehensive and unified evaluation of classical and deep learning–based machine learning techniques for intelligent computer applications. Support Vector Machines, Random
Forests, Gradient Boosting Machines, Artificial Neural Networks,Convolutional Neural Networks, and Long Short-Term Memory networks are implemented and evaluated using benchmark datasets representing classification and regression tasks. Model performance is assessed using accuracy, precision, recall, F1-score, mean absolute error, root mean square error, and coefficient of determination, along with computational indicators including Interdisciplinary Engineering and Technology Management
training time, inference latency, and memory usage. The results demonstrate that Convolutional Neural Networks achieve the highest classification accuracy of 97.1 %, representing an improvement of approximately 5.7 % over classical models, while Support Vector Machines exhibit superior computational efficiency. The study establishes performance–efficiency trade-offs and provides practical
design guidelines for deploying scalable and resource-aware
intelligent computer applications.