Intelligent Systems and Machine Learning

Current Projects

We are developing a variety of interdisciplinary research proposals and have several projects within the following main themes. You can see the current research projects from the web pages of the faculty team members.

Student Projects supervised by Dr. Maiga Chang

Ted Krahn (Personalized Study Guide Generation based on Behaviour Analytics):

This ongoing research has designed and developed two Moodle plug-ins, Behaviour Analytics (https://moodle.org/plugins/block_behaviour) and LORD (Learning Object Relation Discovery, https://moodle.org/plugins/block_lord). The LORD plug-in can measure the similarity between two pieces of learning materials according to their content in English with the help of WordNet and Natural Language Processing so a reasonable and objective Learning Object Graph (LOG) can be created to represent the sequential learning behaviours that students may have among learning materials. The LORD plug-in not only supports learning materials written in English but also French and Hindi with the Word-Sentence Natural Language Processing service (http://ws-nlp.vipresearch.ca/). The Behaviour Analytics plug-in clusters students into different groups based on students' learning behaviour pattern extracted from the log and provides teachers visualization of their students' learning behaviour. Teachers can see how students walk through from one learning object to another; review the common pattern that students in the same group have; and, annotate a group's common pattern or an individual student's pattern based on their own observations. The ongoing research is going to create personalized study guide for individual students according to the results of connecting their extracted learning behaviour patters to various learning styles.

Rip Pennell (Automated Spoken Language Detection):

This ongoing research adopts i-vector algorithm to train neural networks so the trained model can listen to an audio segment and classify a language being spoken in real-world scenarios between multiple languages. The trained model currently is capable of identifying seven languages (English, French, Russian, Chinese used in Mainland China, Arabic, Persian, and German) in acceptable accuracy and has been embeded into a mobile app ready for the further usability study. This research poses as an important and essential base to seamless communication between individuals from all regions of the world.

Miao-Han Chang (Annotation Behaviour Clustering):

The research has designed and developed a platform called GRACE (General Rapid Annotation Clustering Enhancement). Teachers can create online reading activities for their students on GRACE and students can do various annotations on the text freely. Moreover, with the help of a bio-inspired innovative clustering method students can see annotation recommendations whenever they make an annotation. The recommendations are found based on the similarity that the students' annotations have from other students made on the same text earlier. The GRACE platform is not only able to help students double check their annotations against with others' and make their annotations more complete and better for exam preparation later, but also can help teachers to figure out what potential learning issue(s) a group of students may have through examining their annotation behaviours.

Student Projects supervised by Dr. Ali Dewan

Detecting student engagement and designing pedagogical intervention in online learning

Online learning has become an important means of delivering distance education. This learning context presents unique challenges, however, leading to a growing demand for real-time engagement detection and pedagogical interventions designed specifically for an online learning environment. Unfortunately, the theories and principles of technology-enabled learning do not yet provide clarity on best practices in developing such functionalities. To our knowledge, no technological system is currently available that can fully sense natural human engagement signals and respond accordingly. We expect our engagement assessment system combined with pedagogical agents embedded within the online learning environment will fill this gap. We seek to design, implement, and evaluate a system that uses emotion sensing to guide pedagogical responses that will (a) enhance engagement by redirecting the attention of students to their learning activities, and (b) provide students with realistic and effective learning experiences in online courses. This system will be sensitive to the engagement of learners, with the aim of fostering a successful online learning experience. We will implement the system in an online learning environment (e.g., Moodle), where we will perform empirical studies and develop prototype applications to test and explore theoretical concepts. We are most interested in these questions: In what ways and to what extent can online learners’ learning behaviours be properly recorded and recognized? What can these behaviours reveal about the engagement level of individual learners? How can we best respond to learner engagement cues using a pedagogical agent to control the intervention?

Student-Facing Educational Dashboard Design for Online Learners co-supervised by Dr Oscar Lin

The current shift from traditional classrooms to online learning in higher education calls for more attention to self-regulated learning. This research is motivated by the growing interest in potential of using learning analytics dashboard (LAD) to increase individuals’ self-regulation by creating visibility into their performance in various applications. This study explores how data visualization can be integrated with online learning to improve learners’ performance through enhancing their skills in planning and organization. We are working on the design of a comprehensive LAD, focusing on micro-level of learning analytics to support learning activities of students. The LAD includes the following two features to enhance students’ self-regulation in online learning: (1) a function to track students’ progress compared to other students’ over time; (2) reminders to help students with upcoming deadlines and auto–generating to do lists. The hypothesis is that the LAD will increase students’ engagement and motivate self-regulation in an online learning environment. This study is significant because it contributes to the body of knowledge by exploring how student-generated data can be used to improve self-regulated learning. The practical contribution of this study is to create a personalized LAD for students based on the learner-generated data to benefit students’ organization skill, planning skill, and motivation.

Student Projects supervised by Dr. Dunwei Wen

Mr. Robert Ellis, MSc IS Thesis title, “Leveraging Local and Global Word Context for Multi-Label Document Classification”, completed

Mr. Muhammad Sheikh, MSc IS Project title, “Sentiment Analysis Using Deep Neural Networks”, completed.

Student Projects supervised by Dr. Oscar Lin

Mr. Kishan Komaravolu, MSc IS Thesis title, “Adaptive Formative Assessment using FTRL algorithm”, In-progress, co-supervised by Dr Ali Dewan

Mr. John Mui, MSc IS Essay title, “A survey on Multi-armed bandit algorithms for adaptive learning”, completed.