Mathematical algorithms trained on historical data to identify complex patterns and make predictions without explicit programming. In industrial engineering, they are crucial for predicting equipment failures, optimizing thermodynamic processes, and reducing energy consumption.