With the rapid expansion of cryptocurrency exchanges and cloud-based platforms, cybersecurity threats have become increasingly sophisticated. Traditional Intrusion Detection Systems (IDS), which rely on signature-based detection, struggle to identify novel and zero-day attacks. To address these challenges, this research proposes a hybrid intrusion detection framework that combines Artificial Neural Networks (ANN) for supervised classification of known attacks and DBSCAN (Density-Based Spatial Clustering of Applications with Noise) for unsupervised anomaly detection of unknown threats. Moreover, SHAP (SHapley Additive Explanations) is added to amplify model interpretability, so that security experts are able to efficiently interpret predictions. The above framework is analyzed using a benchmark Network Intrusion Detection System (NIDS) data-set, modeled after realistic attack scenarios. The ANN submodule attains top-classification accuracy, capturing established cyber threats with an F1-score of 0.98, and DBSCAN achieves effective discovery of new-type anomalies by getting a silhouette value of 0.8857. Explorability in association with inclusion is made feasible with a better insight into the primary network characteristics that contribute to anomaly detection. Leveraging supervised and unsupervised learning, this hybrid technique improves the capacity of IDS to identify changing cyber threats in cryptocurrency markets and cloud networks. The results indicate that combining machine learning with explainability methods dramatically enhances security controls, presenting an effective solution to contemporary cyber defense.