More Like this

1. Student Subtyping via EM-Inverse Reinforcement Learning.

   Yang, X.; Zhou, G.; Taub, M.; Azevedo, R.; & Chi, M. (January 2020, In Proceedings of the 13th International Conference on Educational Data Mining (EDM) 2020.)

   null (Ed.)

   Abstract: In the learning sciences, heterogeneity among students usually leads to different learning strategies or patterns and may require different types of instructional interventions. Therefore, it is important to investigate student subtyping, which is to group students into subtypes based on their learning patterns. Subtyping from complex student learning processes is often challenging because of the information heterogeneity and temporal dynamics. Various inverse reinforcement learning (IRL) algorithms have been successfully employed in many domains for inducing policies from the trajectories and recently has been applied for analyzing students’ temporal logs to identify their domain knowledge patterns. IRL was originally designed to model the data by assuming that all trajectories have a single pattern or strategy. Due to the heterogeneity among students, their strategies can vary greatly and the design of traditional IRL may lead to suboptimal performance. In this paper, we applied a novel expectation-maximization IRL (EM-IRL) to extract heterogeneous learning strategies from sequential data collected from three simulation environments and real-world longitudinal students’ logs. Experiments on simulation environments showed that EM-IRL can successfully identify different policies from the heterogeneous sequences with different strategies. Furthermore, experimental results from our educational dataset showed that EM-IRL can be used to obtain different student subtypes: a “learning-oriented” subtype who learned the material as much as possible regardless of the time in that they spent significantly more time than the other two subtypes and learned significantly; an“efficient-oriented”subtype who learned efficiently in that they not only learned significantly but also spent less time than the first subtype; a “no learning” subtype who spent less amount of time than first subtype and failed to learn. 

   more » « less
2. Using an Observation Protocol To Evaluate Student Argumentation Skills in Introductory Biology Laboratories

   https://doi.org/10.1128/jmbe.00209-22  

   Clevenger, Lindsey; Teshera-Levye, Jennifer; Walker, Joi P.; Vance-Chalcraft, Heather D. (August 2023, Journal of Microbiology & Biology Education)

   Shaffer, Justin (Ed.)

   ABSTRACT Argumentation is vital in the development of scientific knowledge, and students who can argue from evidence and support their claims develop a deeper understanding of science. In this study, the Argument-Driven Inquiry instruction model was implemented in a two-semester sequence of introductory biology laboratories. Student’s scientific argumentation sessions were video recorded and analyzed using the Assessment of Scientific Argumentation in the Classroom observation protocol. This protocol separates argumentation into three subcategories: cognitive (how the group develops understanding), epistemic (how consistent the group’s process is with the culture of science), and social (how the group members interact with each other). We asked whether students are equally skilled in all subcategories of argumentation and how students’ argumentation skills differ based on lab exercise and course. Students scored significantly higher on the social than the cognitive and epistemic subcategories of argumentation. Total argumentation scores were significantly different between the two focal investigations in Biology Laboratory I but not between the two focal investigations in Biology Laboratory II. Therefore, student argumentation skills were not consistent across content; the design of the lab exercises and their implementation impacted the level of argumentation that occurred. These results will ultimately aid in the development and expansion of Argument-Driven Inquiry instructional models, with the goal of further enhancing students’ scientific argumentation skills and understanding of science. 

   more » « less
3. Using Social Network Analysis to Study the Social Structures of Inclusion

   Pearson, Nelson; Major, Justin; Godwin, Allison; Kirn, Adam (January 2018, ASEE annual conference & exposition)

   The purpose of this research paper is to understand how diverse students are incorporated into the social structure of a large enrollment first-year engineering design course. Despite previous work demonstrating the benefits of diverse individuals in engineering, little work has examined how diverse students are incorporated into the social networks that exist within engineering classrooms. Social interactions are one of the most influential sources for integration into communities of practice. Through understanding how students interact and the structure of these interactions, we can elucidate how the engineering community includes members of underrepresented populations. Previous social network analysis (SNA) studies have scrutinized student classroom interactions. These studies typically attempt to link classroom interactions to academic outcomes (i.e., grades). In this study, we start to shift the focus away from connecting student interactions to academic outcomes and examine how the structure of student interactions can encourage an inclusive environment in a formal engineering environment. SNA data was collected via an online survey (n = 502, 74% response rate) one month into the semester at a Western land-grant institution. The survey asked first-year engineering students to indicate with whom they had interacted using a pre-populated list of the class roster and open-ended questions. The number of students that were mentioned by a participant (out-degree) is interpreted as a proxy of their sociableness. Whereas, the number of times a student was mentioned by others (in-degree) is interpreted as popularity. We posit that in an inclusive network structure the social behaviors (i.e., in and out-degree) will be independent of students’ demographic characteristics (e.g., race and gender). Nonparametric hypothesis testing (i.e., Kruskal-Wallis and Dunn’s test) was used to investigate the effects of gender and race on both in and out-degree. Results indicate that the social structure of the first-year engineering community is inclusive of both gender and race. Specifically, results indicated no significant differences for in-degree based on measures of race and gender, for students who provided race and gender information. Out-degree was not significantly different based on race. However, women did demonstrate significantly higher out-degree scores (i.e., greater sociableness) than their peers. Building on previous SNA literature, the increased connections expressed by women may lead to increased learning gains or performance within engineering. Results indicated that the social structure of this first-year engineering course, as indicated by in-degree and out-degree, is not significantly different for underrepresented groups. This result begins to illustrate a more complex story than the existing literature has documented of engineering as an unwelcoming environment for underrepresented students. Future work will explore how these structures do or do not persist over time and how individuals develop attitudes towards diverse individuals as a result of these interactions. We hope that the results of this work will provide practical ways to improve engineering climate for underrepresented students. 

   more » « less
4. What Time is It? Student Modeling Needs to Know.

   Mao, Y.; Marwan, S.; Price, T.W.; Barnes, T.; & Chi, M. (January 2020, In Proceedings of the 13th International Conference on Educational Data Mining (EDM) 2020)

   null (Ed.)

   Abstract: Modeling student learning processes is highly complex since it is influenced by many factors such as motivation and learning habits. The high volume of features and tools provided by computer-based learning environments confounds the task of tracking student knowledge even further. Deep Learning models such as Long-Short Term Memory (LSTMs) and classic Markovian models such as Bayesian Knowledge Tracing (BKT) have been successfully applied for student modeling. However, much of this prior work is designed to handle sequences of events with discrete timesteps, rather than considering the continuous aspect of time. Given that time elapsed between successive elements in a student’s trajectory can vary from seconds to days, we applied a Timeaware LSTM (T-LSTM) to model the dynamics of student knowledge state in continuous time. We investigate the effectiveness of T-LSTM on two domains with very different characteristics. One involves an open-ended programming environment where students can self-pace their progress and T-LSTM is compared against LSTM, Recent Temporal Pattern Mining, and the classic Logistic Regression (LR) on the early prediction of student success; the other involves a classic tutor-driven intelligent tutoring system where the tutor scaffolds the student learning step by step and T-LSTM is compared with LSTM, LR, and BKT on the early prediction of student learning gains. Our results show that TLSTM significantly outperforms the other methods on the self-paced, open-ended programming environment; while on the tutor-driven ITS, it ties with LSTM and outperforms both LR and BKT. In other words, while time-irregularity exists in both datasets, T-LSTM works significantly better than other student models when the pace is driven by students. On the other hand, when such irregularity results from the tutor, T-LSTM was not superior to other models but its performance was not hurt either. 

   more » « less
5. Cooperative Knowledge Distillation: A Learner Agnostic Approach

   https://doi.org/10.1609/aaai.v38i13.29322  

   Livanos, Michael; Davidson, Ian; Wong, Stephen (March 2024, Proceedings of the AAAI Conference on Artificial Intelligence)

   Knowledge distillation is a simple but powerful way to transfer knowledge between a teacher model to a student model. Existing work suffers from at least one of the following key limitations in terms of direction and scope of transfer which restrict its use: all knowledge is transferred from teacher to student regardless of whether or not that knowledge is useful, the student is the only one learning in this exchange, and typically distillation transfers knowledge only from a single teacher to a single student. We formulate a novel form of knowledge distillation in which many models can act as both students and teachers which we call cooperative distillation. The models cooperate as follows: a model (the student) identifies specific deficiencies in it's performance and searches for another model (the teacher) who encodes learned knowledge into instructional virtual instances via counterfactual instance generation. Because different models may have different strengths and weaknesses, all models can act as either students or teachers (cooperation) when appropriate and only distill knowledge in areas specific to their strengths (focus). Since counterfactuals as a paradigm are not tied to any specific algorithm, we can use this method to distill knowledge between learners of different architectures, algorithms, and even feature spaces. We demonstrate our approach not only outperforms baselines such as transfer learning, self-supervised learning, and multiple knowledge distillation algorithms on several datasets, but it can also be used in settings where the aforementioned techniques cannot. 

   more » « less