Assume the role of a Senior Data Scientist. I have a dataset with 500 features and high dimensionality, representing time-series sensor data from industrial machinery. Outline a step-by-step strategy to detect anomalies. Your strategy must address the curse of dimensionality, feature selection techniques, and the choice between unsupervised algorithms like Isolation Forest versus autoencoders. Explain how you would minimize false positives in a noisy environment.