Trusting the Explainers: Teacher Validation of Explainable Artificial Intelligence for Course Design

Deep learning models for learning analytics have become increasingly popular over the last few years; however, these approaches are still not widely adopted in real-world settings, likely due to a lack of trust and transparency. In this paper, we tackle this issue by implementing explainable AI methods for black-box neural networks. This work focuses on the context of online and blended learning and the use case of student success prediction models. We use a pairwise study design, enabling us to investigate controlled differences between pairs of courses. Our analyses cover five course pairs that differ in one educationally relevant aspect and two popular instance-based explainable AI methods (LIME and SHAP). We quantitatively compare the distances between the explanations across courses and methods. We then validate the explanations of LIME and SHAP with 26 semi-structured interviews of university-level educators regarding which features they believe contribute most to student success, which explanations they trust most, and how they could transform these insights into actionable course design decisions. Our results show that quantitatively, explainers significantly disagree with each other about what is important, and qualitatively, experts themselves do not agree on which explanations are most trustworthy. All code, extended results, and the interview protocol are provided at https://github.com/epfl-ml4ed/trusting-explainers.

Do Not Trust a Model Because It is Confident: Uncovering and Characterizing Unknown Unknowns to Student Success Predictors in Online-Based Learning

Student success models might be prone to develop weak spots, i.e., examples hard to accurately classify due to insufficient representation during model creation. This weakness is one of the main factors undermining users' trust, since model predictions could for instance lead an instructor to not intervene on a student in need. In this paper, we unveil the need of detecting and characterizing unknown unknowns in student success prediction in order to better understand when models may fail. Unknown unknowns include the students for which the model is highly confident in its predictions, but is actually wrong. Therefore, we cannot solely rely on the model's confidence when evaluating the predictions quality. We first introduce a framework for the identification and characterization of unknown unknowns. We then assess its informativeness on log data collected from flipped courses and online courses using quantitative analyses and interviews with instructors. Our results show that unknown unknowns are a critical issue in this domain and that our framework can be applied to support their detection. The source code is available at https://github.com/epfl-ml4ed/unknown-unknowns.

Ripple: Concept-Based Interpretation for Raw Time Series Models in Education

Time series is the most prevalent form of input data for educational prediction tasks. The vast majority of research using time series data focuses on hand-crafted features, designed by experts for predictive performance and interpretability. However, extracting these features is labor-intensive for humans and computers. In this paper, we propose an approach that utilizes irregular multivariate time series modeling with graph neural networks to achieve comparable or better accuracy with raw time series clickstreams in comparison to hand-crafted features. Furthermore, we extend concept activation vectors for interpretability in raw time series models. We analyze these advances in the education domain, addressing the task of early student performance prediction for downstream targeted interventions and instructional support. Our experimental analysis on 23 MOOCs with millions of combined interactions over six behavioral dimensions show that models designed with our approach can (i) beat state-of-the-art educational time series baselines with no feature extraction and (ii) provide interpretable insights for personalized interventions. Source code: https://github.com/epfl-ml4ed/ripple/.

The More Secure, The Less Equally Usable: Gender and Ethnicity (Un)fairness of Deep Face Recognition along Security Thresholds

Face biometrics are playing a key role in making modern smart city applications more secure and usable. Commonly, the recognition threshold of a face recognition system is adjusted based on the degree of security for the considered use case. The likelihood of a match can be for instance decreased by setting a high threshold in case of a payment transaction verification. Prior work in face recognition has unfortunately showed that error rates are usually higher for certain demographic groups. These disparities have hence brought into question the fairness of systems empowered with face biometrics. In this paper, we investigate the extent to which disparities among demographic groups change under different security levels. Our analysis includes ten face recognition models, three security thresholds, and six demographic groups based on gender and ethnicity. Experiments show that the higher the security of the system is, the higher the disparities in usability among demographic groups are. Compelling unfairness issues hence exist and urge countermeasures in real-world high-stakes environments requiring severe security levels.

Reinforcement Recommendation Reasoning through Knowledge Graphs for Explanation Path Quality

Numerous Knowledge Graphs (KGs) are being created to make recommender systems not only intelligent but also knowledgeable. Reinforcement recommendation reasoning is a recent approach able to model high-order user-product relations, according to the KG. This type of approach makes it possible to extract reasoning paths between the recommended product and already experienced products. These paths can be in turn translated into textual explanations to be provided to the user for a given recommendation. However, none of the existing approaches has investigated user-level properties of a single or a group of reasoning paths. In this paper, we propose a series of quantitative properties that monitor the quality of the reasoning paths, based on recency, popularity, and diversity. We then combine in- and post-processing approaches to optimize for both recommendation quality and reasoning path quality. Experiments on three public data sets show that our approaches significantly increase reasoning path quality according to the proposed properties, while preserving recommendation quality. Source code, data sets, and KGs are available at https://tinyurl.com/bdbfzr4n.

Explaining Bias in Deep Face Recognition via Image Characteristics

In this paper, we propose a novel explanatory framework aimed to provide a better understanding of how face recognition models perform as the underlying data characteristics (protected attributes: gender, ethnicity, age; non-protected attributes: facial hair, makeup, accessories, face orientation and occlusion, image distortion, emotions) on which they are tested change. With our framework, we evaluate ten state-of-the-art face recognition models, comparing their fairness in terms of security and usability on two data sets, involving six groups based on gender and ethnicity. We then analyze the impact of image characteristics on models performance. Our results show that trends appearing in a single-attribute analysis disappear or reverse when multi-attribute groups are considered, and that performance disparities are also related to non-protected attributes. Source code: https://cutt.ly/2XwRLiA.

Generalisable Methods for Early Prediction in Interactive Simulations for Education

Interactive simulations allow students to discover the underlying principles of a scientific phenomenon through their own exploration. Unfortunately, students often struggle to learn effectively in these environments. Classifying students' interaction data in the simulations based on their expected performance has the potential to enable adaptive guidance and consequently improve students' learning. Previous research in this field has mainly focused on a-posteriori analyses or investigations limited to one specific predictive model and simulation. In this paper, we investigate the quality and generalisability of models for an early prediction of conceptual understanding based on clickstream data of students across interactive simulations. We first measure the students' conceptual understanding through their in-task performance. Then, we suggest a novel type of features that, starting from clickstream data, encodes both the state of the simulation and the action performed by the student. We finally propose to feed these features into GRU-based models, with and without attention, for prediction. Experiments on two different simulations and with two different populations show that our proposed models outperform shallow learning baselines and better generalise to different learning environments and populations. The inclusion of attention into the model increases interpretability in terms of effective inquiry. The source code is available on Github (https://github.com/epfl-ml4ed/beerslaw-lab.git).