Technical Guest Lecture held Under IEEE COMSOC Karachi Chapter at
Department of Telecommunication Engineering on Wednesday 08-02-2017.
Title: New Research Trends in the field of Data Analytics and Machine Learning
Guest: Engr. Sander Ali Khowaja, Hankuk University of Foreign Studies, Republic of Korea. (TL Alumni)
Abstract of Talk:
The problem of multi-class classification is considered to be an important issue in machine learning domains, specifically when it is applied to activity recognition systems which deal with multiple classes to recognize a particular activity. One of the strategies to convert a multi-class classification problem into binary sub-problems is the error-correcting output coding (ECOC) method. Most of the time a single classifier is used for such ECOC methods without combining multiple classifiers to solve the multi-class classification problem, which fails to generalize the system on real-time sensor recordings for activity recognition systems. In this paper, we proposed a unique hierarchical classification model based on the combination of two base binary classifiers using selective learning of slacked hierarchy and integrating the training of binary classifiers into a unified objective function. We experimented the proposed method on two publicly available datasets and compared the results with the existing independent base classifiers. Furthermore, the proposed method has also been tested on real-time sensor readings for 3 different subjects to recognize four activities i.e. Walking, Standing, Jogging and Sitting using the trained classification model. The experimental results validate the effectiveness of the proposed method for offline as well as activity recognition systems for real-time sensor readings.