摘要 Abstract 

TinyML技術正帶來新的機會。透過將模型壓縮至資源受限的設備上，TinyML不僅大幅降低了使用功耗和成本，更實現了快速、離線的智慧推理能力。從智慧穿戴裝置健康監測到工業物聯網，TinyML在各個領域展現巨大潛力。本次講座將介紹最佳化演算法的演進，並討論在大規模機器學習時代的研發方向。

TinyML technology is opening up new opportunities. By compressing models to run on resource-constrained devices, TinyML not only significantly reduces power consumption and cost, but also enables fast, offline intelligent inference. From health monitoring in smart wearables to industrial IoT applications, TinyML is demonstrating tremendous potential across various fields. This talk will introduce the evolution of optimization algorithms and explore future research directions in the era of large-scale machine learning.