Activity recognition on mobile device sensor data has been an active research area in mobile and pervasive computing for several years. While the majority of the proposed techniques are based on supervised learning, semi-supervised approaches are being considered to reduce the size of the training set required to initialize the model. These approaches usually apply self-training or active learning to incrementally refine the model, but their effectiveness seems to be limited to a restricted set of physical activities. We claim that the context which surrounds the user (e.g., time, location, proximity to transportation routes) combined with common knowledge about the relationship between context and human activities could be effective in significantly increasing the set of recognized activities including those that are difficult to discriminate only considering inertial sensors, and the highly context-dependent ones. In this paper, we propose CAVIAR, a novel hybrid semi-supervised and knowledge-based system for real-time activity recognition. Our method applies semantic reasoning on context-data to refine the predictions of an incremental classifier. The recognition model is continuously updated using active learning. Results on a real dataset obtained from 26 subjects show the effectiveness of our approach in increasing the recognition rate, extending the number of recognizable activities and, most importantly, reducing the number of queries triggered by active learning. In order to evaluate the impact of context reasoning, we also compare CAVIAR with a purely statistical version, considering features computed on context-data as part of the machine learning process.

多年来，对移动设备传感器数据的活动识别一直是移动和普适计算领域的活跃研究领域。尽管大多数提议的技术都是基于监督学习的，但正在考虑采用半监督方法来减少初始化模型所需的训练集的大小。这些方法通常应用自我训练或主动学习来逐步完善模型，但其效果似乎仅限于一组受限的体育活动。我们主张围绕用户的上下文（例如时间，位置，接近交通路线）与有关背景和人类活动之间关系的常识相结合，可以有效地增加公认的活动集，包括那些仅考虑惯性传感器难以区分的活动以及高度依赖上下文的活动。在本文中，我们提出了一种用于实时活动识别的新型混合半监督和基于知识的系统CAVIAR。我们的方法对上下文数据应用语义推理，以完善增量分类器的预测。使用主动学习不断更新识别模型。从26位受试者获得的真实数据集上的结果表明，我们的方法在提高识别率，扩展可识别活动的数量方面非常有效，最重要的是，减少主动学习触发的查询数量。为了评估上下文推理的影响，我们还将CAVIAR与纯统计版本进行了比较，考虑了在上下文数据中计算出的特征作为机器学习过程的一部分。