Machine learning and artificial intelligence have attracted a lot of attention over the past few decades in many application domains, including Internet of Things (IoT). The adoption of machine learning in IoT systems creates many new opportunities, e.g., detection of health abnormalities using wearable devices, but also involves several major challenges. In particular, IoT systems are decentralized and often limited in terms of computing/energy resources. In this talk, we highlight the key challenges in enabling machine-learning techniques on mobile-health and wearable IoT devices and discuss a few early solutions inspired by the human brain and nervous system.