Multiple-goal recognition from low-level signals

摘要: Researchers and practitioners from both the artificial intelligence pervasive computing communities have been paying increasing attention to task of inferring users' high-level goals low-level sensor readings. A common assumption made by most approaches is that a user either has single goal in mind, or achieves several sequentially. However, real-world environments, often multiple are concurrently carried out, action can serve as step towards goals. In this paper, we formulate multiple-goal recognition problem exemplify it an indoor environment where RF-based wireless network available. We propose goal-recognition algorithm based on dynamic model set show how models evolve over time pre-defined states. Experiments with real data demonstrate our method accurately efficiently recognize interleaving user's trace.