Due to significant advances in sensor technology, studies towards activity recognition have gained interest and maturity in the last few years. Existing machine learning algorithms have demonstrated promising results by classifying activities whose instances have been already seen during training. Activity recognition methods based on real-life settings should cover a growing number of activities in various domains, whereby a significant part of instances will not be present in the training data set. However, to cover all possible activities in advance is a complex and expensive task. Concretely, we need a method that can extend the learning model to detect unseen activities without prior knowledge regarding sensor readings about those previously unseen activities. In this paper, we introduce an approach to leverage sensor data in discovering new unseen activities which were not present in the training set. We show that sensor readings can lead to promising results for zero-shot learning, whereby the necessary knowledge can be transferred from seen to unseen activities by using semantic similarity. The evaluation conducted on two data sets extracted from the well-known CASAS datasets show that the proposed zero-shot learning approach achieves a high performance in recognizing unseen (i.e., not present in the training dataset) new activities.

No abstract

The use of IoT-based Emotion Recognition (ER) systems is in increasing demand in many domains such as active and assisted living (AAL), health care and industry. Combining the emotion and the context in a unified system could enhance the human support scope, but it is currently a challenging task due to the lack of a common interface that is capable to provide such a combination. In this sense, we aim at providing a novel approach based on a modeling language that can be used even by care-givers or non-experts to model human emotion w.r.t. context for human support services. The proposed modeling approach is based on Domain-Specific Modeling Language (DSML) which helps to integrate different IoT data sources in AAL environment. Consequently, it provides a conceptual support level related to the current emotional states of the observed subject. For the evaluation, we show the evaluation of the well-validated System Usability Score (SUS) to prove that the proposed modeling language achieves high performance in terms of usability and learn-ability metrics. Furthermore, we evaluate the performance at runtime of the model instantiation by measuring the execution time using well-known IoT services.

No abstract

Emotions are an essential part of a person’s mental state and influence her/his behavior accordingly. Consequently, emotion recognition and assessment can play an important role in supporting people with ambient assistance systems or clinical treatments. Automation of human emotion recognition and emotion-aware recommender systems are therefore increasingly being researched. In this paper, we first consider the essential aspects of human emotional functioning from the perspective of cognitive psychology and, based on this, we analyze the state of the art in the whole field of work and research to which automated emotion recognition belongs. In this way, we want to complement the already published surveys, which usually refer to only one aspect, with an overall overview of the languages ontologies, datasets, and systems/interfaces to be found in this area. We briefly introduce each of these subsections and discuss related approaches regarding methodology, technology, and publicly accessible artefacts. This comes with an update to recent findings that could not yet be taken into account in previous surveys. The paper is based on an extensive literature search and analysis, in which we also made a particular effort to locate relevant surveys and reviews. The paper closes with a summary of the results and an outlook on open research questions.