Adaptive optimization, also known as profile-guided optimization, is a technique where the optimization space is explored based on the outcome of fitness functions, e.g., the execution time to profile the executable and dynamically modifies/recompiles certain segments of an application under optimization. The profiling provides enough features so the compiler can decide on what portion of the code to be recompiled. An adaptive compiler is placed between a just-in-time (JIT) compiler and an interpreter of instructions. [0]

[0] Ashouri, Amir H., et al. “A survey on compiler autotuning using machine learning.” arXiv preprint arXiv:1801.04405 (2018).