Malware detection is a pivotal challenge in cybersecurity, demanding advanced methods to counter increasingly sophisticated threats. Machine learning has evolved as an enabler in this regard, and powerful detection systems are made available. The paper provides a systematic review of 57 articles that cover the use of machine learning algorithms to detect malware on the computer and, importantly, this is with a picture in mind of mixing the use of signature-based, behavior-based, and hybrid approaches to achieve greater and higher detection rates and the resilience especially against real-world threats. Signature-based methods leverage predefined patterns, behavior-based techniques analyze runtime activities, and hybrid models combine rigid and dynamic assessment for enhanced detection accuracy. The review compares many algorithms, such as classical machine learning (e.g., Support Vector Machine, Naive Bayes, Decision Tree, Random Forest, Logistic Regression, K-Nearest Neighbors) approaches, deep architectures (e.g., Artificial Neural Networks, Convolutional Neural Networks, Recurrent Neural Networks) and advanced models (e.g. Graph Neural Networks, Transformer Models, Vision Transformers). Additionally, hybrid frameworks such as Deep Belief Networks, Transfer Learning Models, Tree Augmented Naive Bayes (TAN), and the Non-Dominated Sorting Genetic Algorithm (NSGA-II) are examined. These methodologies have been applied to diverse datasets, showcasing their applicability in real-world scenarios.