User-dependent aspect model for collaborative activity recognition

摘要: Activity recognition aims to discover one or more users' actions and goals based on sensor readings. In the real world, a single user's data are often insufficient for training an activity model due sparsity problem. This is especially true when we interested in obtaining personalized model. this paper, study how collaboratively use different train that can provide each user. We propose user-dependent aspect collaborative task. Our introduces user variables capture grouping information, so target also benefit from her similar users same group way, greatly reduce need much valuable expensive labeled required capable of incorporating time information handling new recognition. evaluate our real-world WiFi set obtained indoor environment, show proposed outperform several state-of-art baseline algorithms.