Mathematics for AI — Ch.03: Probability [KR]

Probability provides a mathematical framework to describe uncertainty, measure outcome likelihood, and reason about events with incomplete observations, measurement noise, and unpredictable behavior, forming a core foundation for AI. Inference optimization focuses on improving the execution efficiency of trained models during deployment, minimizing latency, optimizing memory utilization, and achieving high throughput with appropriate computational cost across diverse hardware. Memory and I/O subsystems underpin high‑performance AI hardware; actual performance depends on how efficiently data is moved to computational units, with memory capacity, bandwidth, access latency, and communication architecture being key determinants of compute resource utilization.