Student Projects – Autumn 2021

We are offering a range of Master Projects and Semester Projects in data mining and machine learning for (vocational) education ecosystems.


To apply, you are requested to send an e-mail to the contact person, mentioning the topic of interest and attaching your grade transcripts. If you would like to receive more information on our research, do not hesitate to contact us!


Please, note that the list of topics below is not exclusive. In case you have other ideas or proposals, you are welcome to contact a senior member of the lab and talk about possibilities for a tailor-made topic


Machine Learning for Adaptive Teaching Interventions in Flipped Classrooms

Nowadays, flipped classrooms are receiving increasing interest in the instructional design field. This learning strategy requires students to follow pre-class learning activities, before meeting with the teacher and the other peers for an in-person discussion and assessment. For pre-class activities, teachers often provide students with videos and digital content by means of an online learning platform. Engaging with pre-class activities is essential for the success of flipped classrooms, as these activities prepare students for a more active participation in face-to-face lessons.

The goal of student’s projects in this area is to analyze students’ behavior in pre-class activities in flipped courses to enable content personalization and adaptive teaching assistance. These tasks are directly connected with the current research of the lab involving the design, implementation, and deployment of success prediction systems that, by anticipating the academic performance (passing/failing or the grade) of a student in a course, serve as a means for timely interventions. The underlying data set includes clickstreams left by students while interacting with content (e.g., videos), peers (e.g., forums), and assessments (e.g., quizzes) in EPFL flipped courses delivered through the EPFL Courseware platform, such as Linear Algebra and Functional Programming. Student’s projects in this area will involve both technical and pedagogical aspects needed to model time series and to define education-related indicators of student’s success in pre-class activities. Machine learning models with these indicators as features will be then developed, evaluated, and discussed, especially in terms of performance, interpretability, and impact in the flipped context.

For more details on the project, please do not hesitate to contact us!

Keywords: Flipped Classrooms, Clickstream Analysis, Early Warning Systems, Success Prediction.

Preferred Skills: Proficiency in Python; Experience with Machine Learning Algorithms.

Useful Tools: Jupyter Notebooks, Numpy, Pandas, ScikitLearn, Seaborn, TensorFlow.

Contact: Dr. Mirko Marras (Website: mirkomarras.com / E-mail:
[email protected]).


Machine Learning for Massive Open Online Course Platforms

Over the last years, Massive Open Online Course (MOOC) platforms have been providing life-changing educational opportunities to millions of individuals, with unrestricted participation to any course of their choice, anytime, anywhere. Notable examples include Udemy, Coursera, edX, and Udacity. Top-tier universities, such as EPFL, are now offering a wide range of MOOCs as well, providing students with a way to learn in a setting similar to an online class, but with a loosely structured schedule and the opportunity of interacting with a large number of peers from around the world. However, scaling up education online towards these numbers is presenting core challenges, such as hardly manageable classes and overwhelming content alternatives.

The goal of students’ projects in this area is to analyze how individuals interact in MOOC platforms and provide them with timely supporting services. These activities are closely related with the lab research ranging from personalized recommendation (e.g., resources you might be interested in or peers who may have similar interests) to student’s dropout/success prediction aimed at anticipating whether and why a student will not complete (pass) a course, as examples. The data set includes millions of clickstreams records left by students from all over the world in hundreds of MOOCs offered by EPFL professors on platforms like Coursera, edX, and Courseware. Both technical and pedagogical elements will be combined in student’s projects in this area, to relate clickstream behavior with numerical indicators pertaining to student’s learning aspects. It will be inspected the extent to which these indicators can serve as a predictive feature when fed into the related machine learning models and how this predictivity can transfer/generalize across courses. These models will be evaluated, discussed, paying attention to performance, interpretability, and impact.

For more details on the project, please do not hesitate to contact us!

Keywords: Neural Networks, Clickstream Analysis, Dropout Prediction, Recommender Systems.

Preferred Skills: Proficiency in Python; Experience with Machine Learning Algorithms.

Useful Tools: Jupyter Notebooks, Scikit-Learn, Tensorflow, SpaCy, NLTK.

Contact: Dr. Mirko Marras (Website: mirkomarras.com / E-mail: [email protected])


Responsible Machine Learning for Education

Educational technologies and platforms are increasingly integrating predictive models to provide data-driven support to students, instructors, and other educational stakeholders, with machine learning often being their core part of these predictive models. For instance, cognitive tutors help students mastery skills by adaptively providing them with learning materials, and student support systems assist or flag students based on how likely they may disengage, fail an exam, or experience certain affective states. Other applications include early dropout and course recommendations, as examples. As these predictive models play a relevant role in the educational landscape, it becomes essential to explore how they deal with beyond-accuracy aspects like ethics, fairness, transparency, explainability, and accountability. The laboratory is now interested in fairness and explainability in students’ success models for MOOC and flipped courses (other systems are under consideration).

  • Fairness Track. The goal of student’s projects in this track is to assess algorithmic unfairness in machine-learning models deployed in education and design strategies to ensure that models provide fairer treatments across students belonging to protected groups, while retaining model performance. To this end, several measures (e.g., equal opportunity, equalized odds) have been defined in the generic machine-learning domain to assess how fair machine-learning models are. Student’s projects will be devoted to the analysis of these fairness measures in the context of EPFL MOOC and flipped courses clickstream data and models and the formulation of new fairness definitions and unfairness mitigation procedures tailored to the educational domain.
  • Explainability Track. The goal of student’s projects in this track is to analyze strategies for making predictions of machine-learning models deployed in education more explainable and transparent, allowing teachers and students to better react to model’s decisions. To this end, several methods (e.g., LIME, SHAP, Shapley Values) have been investigated in the generic machine-learning domain to give explanations alongside predictions from black-box machine-learning models. Student’s projects will be devoted to the analysis of these explainability algorithms in the context of EPFL MOOC clickstream data and models, the comparison of their strengths and weaknesses, and the design of variants tailored to the educational domain.

For more details on the project, please do not hesitate to contact us!

Keywords: Deep Learning, Natural Language Processing (NLP), Natural Language Generation (NLG), Interpretable AI, Fairness, Bias.

Preferred Skills: Good Background in Statistics; Proficiency in Python; Experience with Machine Learning Algorithms; Proficiency in Natural Language Processing.

Useful Tools: Jupyter Notebooks, Scikit-Learn, Tensorflow, PyTorch, SpaCy, NLTK, SQL.

Contacts:


Interactive Simulations for Vocational Education

 

 

Interactive simulations can often be a helpful learning tool, providing a safe environment in which learners can freely explore and experiment without the dangers, risks and constraints of real-world setups. When appropriately utilized by teachers and students, such tools can especially facilitate the understanding of concepts that are normally difficult to grasp. The instantaneous feedback provided, such as visualizations and animations, can support students’ comprehension by making mechanisms visible that otherwise would remain hidden. Another alternative having the goal to innovate traditional instructional approaches is represented by learning games (or serious games). Through gamification, these applications aim at conveying learning content in a more playful way, capitalizing on the idea that such approaches may have a positive impact on learning motivation and engagement.

The vocational training for many professions could benefit from the targeted use of interactive simulations and/or learning games. At this moment, the laboratory is interested in developing and analyzing user interactions with chemistry simulations for lab technicians, thermal simulations for heating technicians, simulations for electronics technicians using mobile robots, learning games for vocational education in the global south. For more details on the projects, please do not hesitate to contact us!

Keywords: Interactive Simulations, Learning Games, Gamification

Preferred skills: Experience with programming; Interest for digital games; Experience with web/app development can be helpful.

Useful tools: Unity, HTML5, Android Studio

Contact: Dr. Christian Giang (E-mail: [email protected])


Digital Game-based Learning

Another alternative having the goal to innovate traditional instructional approaches is represented by digital game-based learning. Through gamification, these applications aim at conveying learning content in a more playful way, capitalizing on the idea that such approaches may have a positive impact on learning motivation and engagement.

At this moment, the laboratory is interested in designing and testing new digital game-based learning approaches such as **digital escape rooms and scenario-based learning games, as well as in analyzing users’ behaviors in such environments. Current projects involve e.g. the development of a digital escape room activity for the vocational training of pharma assistants or a mobile learning game for vocational education in the global south.

Student projects in this field will naturally involve many technical aspects (coding, data analysis etc.). However, they usually also involve educational game design as well as conducting user studies (also in classrooms). For more details on the projects, please do not hesitate to contact us!

Keywords: Digital Game-based Learning, Educational Game Design

Preferred skills: Experience with programming; Experience with web/app development can be helpful; Interest for game design; Interest for education; German and French can be helpful (for classroom studies)

Useful tools:

  • Development: Unity, HTML5, Android Studio, JavaScript, PHP, MySQL
  • Analysis: Python, Jupyter Notebooks, Pandas, Scikit-learn

Contact: Dr. Christian Giang (E-mail: [email protected])