论文部分内容阅读
针对智能环境中活动模式的学习和挖掘花销大、难以实际操作等问题,提出了能够有效地将已有活动模式迁移到新环境的整体框架。迁移学习框架将活动模式的迁移过程分解为轨迹的迁移和触发持续时间的迁移,首先对已有活动模式中的活动轨迹以及触发持续时间模糊化;然后采用备选轨迹生成(ATSG)算法在新环境中生成备选轨迹集;最后采用相似度计算(SC)算法进行活动模式中的轨迹与备选轨迹间的匹配,利用活动轨迹映射(TM)算法和触发持续时间迁移(TDT)算法对活动信息进行迁移,从而在新环境中得到活动模式。理论分析和实验结果表明,相比于基于频繁模式挖掘得到活动模式的方法,本文方法大幅度地降低了得到活动模式所需的时间开销,同时,利用本文方法获取的活动模式取得了较好的活动识别效果。
In order to solve the problems that the learning and mining of activity patterns in smart environment are costly and hard to be practiced, a holistic framework that can effectively transfer the existing activity patterns to the new environment is proposed. The framework of migration learning decomposes the migration process of activity patterns into trajectory migration and migration duration triggering. Firstly, the activity trajectory and the duration of triggering in the existing activity patterns are obscured; then, the alternative trajectory generation (ATSG) Environment. In the end, the similarity calculation (SC) algorithm is used to match the trajectories in the active mode and the candidate trajectories, and the activity trajectory mapping (TM) algorithm and the trigger duration migration (TDT) Information is migrated to get a model of activity in the new environment. Theoretical analysis and experimental results show that compared with the method based on frequent pattern mining, the method of this paper drastically reduces the time cost of obtaining the activity pattern. At the same time, the activity pattern obtained by this method achieves better Activity recognition effect.