An official website of the United States government
There is a long history of research on the development of models to detect and study student behavior and affect. Developing computer-based models has allowed the study of learning constructs at fine levels of granularity and over long periods of time. For many years, these models were developed using features based on previous educational research from the raw log data. More recently, however, the application of deep learning models has often skipped this feature-engineering step by allowing the algorithm to learn features from the fine-grained raw log data. As many of these deep learning models have led to promising results, researchers have asked which situations may lead to machine-learned features performing better than expert-generated features. This work addresses this question by comparing the use of machine-learned and expert-engineered features for three previously-developed models of student affect, off-task behavior, and gaming the system. In addition, we propose a third feature-engineering method that combines expert features with machine learning to explore the strengths and weaknesses of these approaches to build detectors of student affect and unproductive behaviors.
more »
« less
Machine-Learned or Expert-Engineered Features? Exploring Feature Engineering Methods in Detectors of Student Behavior and Affect
There is a long history of research on the development of models to detect and study student behavior and affect. Developing computer-based models has allowed the study of learning constructs at fine levels of granularity and over long periods of time. For many years, these models were developed using features based on previous educational research from the raw log data. More recently, however, the application of deep learning models has often skipped this feature engineering step by allowing the algorithm to learn features from the fine-grained raw log data. As many of these deep learning models have led to promising results, researchers have asked which situations may lead to machine-learned features performing better than expert-generated features. This work addresses this question by comparing the use of machine-learned and expert-engineered features for three previously-developed models of student affect, off-task behavior, and gaming the system. In addition, we propose a third feature-engineering method that combines expert features with machine learning to explore the strengths and weaknesses of these approaches to build detectors of student affect and unproductive behaviors.
more »
« less
- Award ID(s):
- 1724889
- PAR ID:
- 10157372
- Date Published:
- Journal Name:
- Proceedings of the Twelfth International Conference on Educational Data Mining
- Page Range / eLocation ID:
- 508-511
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Training machine learning (ML) models for scientific problems is often challenging due to limited observation data. To overcome this challenge, prior works commonly pre-train ML models using simulated data before having them fine-tuned with small real data. Despite the promise shown in initial research across different domains, these methods cannot ensure improved performance after fine-tuning because (i) they are not designed for extracting generalizable physics-aware features during pre-training, (ii) the features learned from pre-training can be distorted by the fine-tuning process. In this paper, we propose a new learning method for extracting, preserving, and adapting physics-aware features. We build a knowledge-guided neural network (KGNN) model based on known dependencies amongst physical variables, which facilitate extracting physics-aware feature representation from simulated data. Then we fine-tune this model by alternately updating the encoder and decoder of the KGNN model to enhance the prediction while preserving the physics-aware features learned through pre-training. We further propose to adapt the model to new testing scenarios via a teacher-student learning framework based on the model uncertainty. The results demonstrate that the proposed method outperforms many baselines by a good margin, even using sparse training data or under out-of-sample testing scenarios.more » « less
-
Training machine learning (ML) models for scientific problems is often challenging due to limited observation data. To overcome this challenge, prior works commonly pre-train ML models using simulated data before having them fine-tuned with small real data. Despite the promise shown in initial research across different domains, these methods cannot ensure improved performance after fine-tuning because (i) they are not designed for extracting generalizable physics-aware features during pre-training, (ii) the features learned from pre-training can be distorted by the fine-tuning process. In this paper, we propose a new learning method for extracting, preserving, and adapting physics-aware features. We build a knowledge-guided neural network (KGNN) model based on known dependencies amongst physical variables, which facilitate extracting physics-aware feature representation from simulated data. Then we fine-tune this model by alternately updating the encoder and decoder of the KGNN model to enhance the prediction while preserving the physics-aware features learned through pre-training. We further propose to adapt the model to new testing scenarios via a teacher-student learning framework based on the model uncertainty. The results demonstrate that the proposed method outperforms many baselines by a good margin, even using sparse training data or under out-of-sample testing scenarios.more » « less
-
Automatic pain intensity assessment from physiological signals has become an appealing approach, but it remains a largely unexplored research topic. Most studies have used machine learning approaches built on carefully designed features based on the domain knowledge available in the literature on the time series of physiological signals. However, a deep learning framework can automate the feature engineering step, enabling the model to directly deal with the raw input signals for real-time pain monitoring. We investigated a personalized Bidirectional Long short-term memory Recurrent Neural Networks (BiLSTM RNN), and an ensemble of BiLSTM RNN and Extreme Gradient Boosting Decision Trees (XGB) for four-category pain intensity classification. We recorded Electrodermal Activity (EDA) signals from 29 subjects during the cold pressor test. We decomposed EDA signals into tonic and phasic components and augmented them to original signals. The BiLSTM-XGB model outperformed the BiLSTM classification performance and achieved an average F1-score of 0.81 and an Area Under the Receiver Operating Characteristic curve (AUROC) of 0.93 over four pain states: no pain, low pain, medium pain, and high pain. We also explored a concatenation of the deep-learning feature representations and a set of fourteen knowledge-based features extracted from EDA signals. The XGB model trained on this fused feature set showed better performance than when it was trained on component feature sets individually. This study showed that deep learning could let us go beyond expert knowledge and benefit from the generated deep representations of physiological signals for pain assessment.more » « less
-
Research into "gaming the system" behavior in intelligent tutoring systems (ITS) has been around for almost two decades, and detection has been developed for many ITSs. Machine learning models can detect this behavior in both real-time and in historical data. However, intelligent tutoring system designs often change over time, in terms of the design of the student interface, assessment models, and data collection log schemas. Can gaming detectors still be trusted, a decade or more after they are developed? In this research, we evaluate the robustness/degradation of gaming detectors when trained on old data logs and evaluated on current data logs. We demonstrate that some machine learning models developed using past data are still able to predict gaming behavior from student data collected 16 years later, but that there is considerable variance in how well different algorithms perform over time. We demonstrate that a classic decision tree algorithm maintained its performance while more contemporary algorithms struggled to transfer to new data, even though they exhibited better performance on both new and old data alone. Examining the feature importances provides some explanation for the differences in performance between models, and offers some insight into how we might safeguard against detector rot over time.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
