Issue link: https://resources.mouser.com/i/1481902
The Rise of Machine Learning for Edge-AI Devices Louis Gobin, STMicroelectronics Intelligence at the Edge 4 A rtificial Intelligence (AI) is the broad field of computer science that incorporates human-like intelligence into machines. Machine Learning (ML) is the science of training machines to make decisions without explicit programming. Deep Learning is a subfield of ML, based on "neural network" algorithms that mimic the decision-making ability of the human brain. Over the past decade ML has evolved beyond the domain of university labs and research centers focusing on business applications (Figure 1). Today, it is broadly applied across many industries. People encounter ML throughout the day with smartphone face detection, GPS routing, email spam filtering, language translation, credit card transaction security, personalized web ads, virtual assistants, facial recognition, and disease diagnoses. It is therefore not surprising that ML ambitions are high. Multiple recent surveys indicate that most business executives and managers believe ML will allow their companies to gain a competitive market advantage. Yet many companies have struggled to create customer value with ML technology (Figure 2). The triumphs of AI have come mostly from giant tech companies with vast data and cloud resources. But Cloud-AI is complex and challenging for companies with modest expertise, resources, and data access. Production deployments of ML models with Cloud-AI have therefore been challenging for many of them. Fortunately, this situation is rapidly improving with Edge-AI adoption, where ML algorithms are executed directly on handheld and embedded devices without cloud dependency.