Assignments

This assignment focuses on the analysis of an Intelligent Tutoring System (ITS) selected from a set of review papers shared in the course folder. The papers include multiple ITS implementations, some explicitly described and others requiring deeper examination. Each student is required to select one ITS, register the choice in the shared sheet to avoid duplication, and submit a 3-page document detailing the system’s domain, pedagogical, and learner models, along with a summary of the research study conducted and its results. The submission should also include a critical argument on the effectiveness of the ITS for learning. The assignment is due on 14th January, carries a 5% weightage, and must be submitted via Moodle.

This assignment involves designing and implementing a short instructional learning module using both GIFT Tutoring and Articulate Rise, followed by a pedagogical comparison of the two approaches. Students will work in groups of three to select a simple, well-defined topic that can be taught to a novice learner within 15 minutes; topics must be unique across groups to avoid grade penalties. The learning content (video or PDF) should be clear, concise, and aligned with the learning objective, and must be implemented in both platforms to enable comparison. Each group will also design a non-open-ended assessment (e.g., MCQs) that produces a measurable score between 0 and 1 and aligns with the learning objective. The submission includes links/files for both implementations and a 3-page written comparison evaluating which instructional model or framework is better suited for the topic, learner experience, and assessment strategy, supported by pedagogical reasoning. The assignment is to be completed in groups of three and submitted by February 13, with all group members required to submit the same files.

This assignment focuses on emotion and affect detection using machine learning models. Students will work in groups of three to first create a labeled dataset by recording videos that capture 4–5 distinct emotions, generating a minimum of 3000 images per emotion, and fine-tuning an emotion detection model to evaluate its performance. In the second stage, the same dataset will be used to apply at least two affect detection approaches (such as circumplex models) to estimate valence and arousal values and identify affective states, supported by an in-depth analysis of the methods described in the provided research papers. The submission must include a Colab/Jupyter Notebook and a 3-page summary describing the model training process and a comparative evaluation of the affect detection models and their performance. The assignment is due on March 15th.