Author: Ankit Verma, Anupma Kulkarni, Ravi Shankar Mishra
Abstract:The rapid proliferation of mobile and edge devices has necessitated the shift of machine learning (ML) models from cloud-based systems to on-device implementations. On-device machine learning provides several advantages, including reduced latency, increased privacy, and offline functionality. TensorFlow Lite (TFLite), a lightweight version of TensorFlow, is specifically designed for deploying ML models on resource-constrained devices such as smartphones, embedded systems, and IoT devices. This review paper explores the integration of TensorFlow Lite for on-device ML applications, discussing its architecture, optimization techniques, deployment strategies, and real-world use cases. Furthermore, the challenges and future directions in on-device ML adoption are analyzed. The paper includes practical examples, performance comparison tables, and an evaluation of TensorFlow Lite’s effectiveness in diverse scenarios.
Keywords: On-device machine learning, TensorFlow Lite, mobile AI, edge computing, model optimization, embedded ML
Full Issue
| View or download the full issue | 24-37 PDF |