Unsupervised Learning Approaches for Non-Stationary Data Streams

Abstract

Modern society is surrounded by several applications which are daily generating large volumes of data. Nowadays, anyone can monitor their physical activities in real-time by using smartphones or wearable devices. Also, business and governments can learn more about their clients and citizens by analysing information from social media, for example. This data is called data streams when it is a sequence of data generated continuously, usually at high speed. This data is also potentially unbounded in size and may not be strictly stationary.Extracting useful knowledge from data streams is challenged due to several constraints. A data stream requires that a learning algorithm acts in dynamic environments. Meaning that the learning algorithm should allow for real-time processing. Moreover, it should be able to adapt to changes over time, considering the non-stationary nature of the data stream.Most of this thesis focus on clustering approaches. However, given the popularity of neural networks, we also propose Ensemble of Auto-Encoders. This approach is based on the combination of auto-encoders into an ensemble model. Each auto-encoder is specialised on recognising one particular class. The Ensemble of Auto-Encoders has a modular structure that has the advantage of making the model easily adapted to the changes from the data. Besides, it allows for personalised models because the model can adapt to the most request classes. This contribution is applied to the problem of Human Activity Recognition. Experimental results show the potential of the approaches mentioned.