inference
also: ML inference, model inference, prediction
Inference in Linux typically refers to the process of a machine learning model making predictions on new data, or more broadly, the execution of a trained AI/ML model to generate outputs from inputs without further training.
In the context of Linux and modern software development, inference most commonly refers to the runtime execution phase of machine learning models. Once a model has been trained (a separate process), inference is when that model processes new, unseen data to generate predictions or classifications.
For example, a trained image recognition model running on a Linux server might perform inference on uploaded photos to identify objects in them. Tools like TensorFlow, PyTorch, and ONNX Runtime on Linux handle inference workloads, often leveraging GPU acceleration for performance.
In broader computing contexts, inference can also mean the process of deriving conclusions or making educated guesses based on available information—such as a program inferring a file type from its extension or content.