Feature Learning using Convolutional Denoising Autoencoder for Activity Recognition

The hand-crafted (time and frequency-based) features that are extracted and selected are heuristic and rely on expert knowledge of the domain. The features may be effective in certain specific settings, but the same features might fail to discriminate the activities in a more general environment. Furthermore, hand-crafted feature extraction and selection are time-consuming, laborious and…

A Unified Generative Model using Generative Adversarial Network for Activity Recognition

One of the important factors of high accuracy of deep learning is the sufficient amount of training data. A robust and reliable model needs a vast amount of data to precisely capture the underlying pattern of the data. However, data collection in activity recognition is expensive and time-consuming especially from the elderly people. In this…

Ontology-based sensor fusion for activity recognition

Context-aware activity recognition systems are dealing with heterogeneous sensors and these sensors are providing data at different sampling rate and output forms. Wearable sensors such as accelerometers and gyroscopes provide fast and real-time raw data which has to be interpreted before being useful to the application. Whereas ambient sensors such as temperature, humidity or object-interaction…

UoA IELAB ADL v2

UoA Intelligent Environment Laboratory (IELAB) ADL Normal Testbed v2

Adaptive Sliding Window for Physical Activity Recognition

A sliding window with a fixed size is not an effective approach for activity recognition system. Misclassifications could still happen especially for transitional activities. This is due to the fact that the length of transitional activity signals varies depending on the time to complete the activity [1], [2]. To overcome the problem, the window size…

Enhancing ontological reasoning with uncertainty handling for activity recognition

Handling uncertainty is a challenge in activity recognition. Uncertainty can be due to sensor errors (e.g. run out of batteries, imprecise outputs, missing activations etc.) and communication failures. and variability in human activities. These issues may significantly influence the accuracy of activity recognition. Data-driven approaches use machine learning techniques such as Decision Tree, na├»ve Bayes…

Evaluation of accelerometer based multi-sensor versus single-sensor activity recognition systems

The paper describes the comparison of activity recognition system consisting of a single accelerometer sensor and multiple accelerometer sensors. The paper also presented the influence of sampling rate towards recognition accuracies, most widely used features in this research area and the time take to calculate them, and comparison between classifiers in activity recognition. The result…