This chapter briefly overviews robust machine-learning solutions for wearable IoT applications. Furthermore, it presents one of the earliest attempts in presenting an autonomous learning framework for wearables. The focus, in particular, is on cases where a new sensor is added to the system and the new (untrained) sensor is worn/used on various body locations. The process of autonomous learning automatically leads to a new collaborative decision-making algorithm. Addressing the problem of expanding pattern-recognition capabilities from a single setting algorithm with a predefined configuration to a dynamic setting where sensors can be added, displaced, and used unobtrusively is challenging. In such cases, successful knowledge transfer is needed to improve the learning performance by avoiding expensive data collection and labeling efforts. In this chapter, a novel and generic approach to transfer learning capabilities of an existing static sensor to a newly added dynamic sensor is described.

Chapter Contents:

• 4.1 Transfer learning in wearable IoT
• 4.2 Synchronous dynamic view learning
• 4.2.1 Problem definition
• 4.2.2 Problem formulation
• 4.2.3 Overview of autonomous learning
• 4.3 Minimum disagreement labeling
• 4.3.1 Label refinement
• 4.4 Experimental analysis
• 4.4.1 Evaluation methodology
• 4.4.2 Accuracy of transferred labels
• 4.4.3 Accuracy of activity recognition
• 4.4.4 Precision, recall, and F1-measure
• 4.5 Summary
• References

Inspec keywords: body sensor networks; learning (artificial intelligence); Internet of Things; decision making

Other keywords: dynamic sensor; wearable IoT applications; autonomous learning framework; knowledge transfer; static sensor; pattern-recognition capabilities; autonomous collaborative learning; robust machine-learning solutions; collaborative decision-making algorithm

Subjects: Computer networks and techniques; Computer communications; Knowledge engineering techniques; Wireless sensor networks