More Like this

1. Automatic Interpretable Personalized Learning

   Prihar, Ethan ; Haim, Aaron ; Sales, Adam ; Heffernan, Neil ( June 2022 , Learning at Scale)

   Personalized learning stems from the idea that students benefit from instructional material tailored to their needs. Many online learning platforms purport to implement some form of personalized learning, often through on-demand tutoring or self-paced instruction, but to our knowledge none have a way to automatically explore for specific opportunities to personalize students’ education nor a transparent way to identify the effects of personalization on specific groups of students. In this work we present the Automatic Personalized Learning Service (APLS). The APLS uses multi-armed bandit algorithms to recommend the most effective support to each student that requests assistance when completing their online work, and is currently used by ASSISTments, an online learning platform. The first empirical study of the APLS found that Beta-Bernoulli Thompson Sampling, a popular and effective multi-armed bandit algorithm, was only slightly more capable of selecting helpful support than randomly selecting from the relevant support options. Therefore, we also present Decision Tree Thompson Sampling (DTTS), a novel contextual multi-armed bandit algorithm that integrates the transparency and interpretability of decision trees into Thomson sampling. In simulation, DTTS overcame the challenges of recommending support within an online learning platform and was able to increase students’ learning by as much as 10% more than the current algorithm used by the APLS. We demonstrate that DTTS is able to identify qualitative interactions that not only help determine the most effective support for students, but that also generalize well to new students, problems, and support content. The APLS using DTTS is now being deployed at scale within ASSISTments and is a promising tool for all educational learning platforms. 

   more » « less
2. Automatic Interpretable Personalized Learning

   Prihar, Ethan ; Haim, Aaron' Sales ; Heffernan, Neil ( June 2022 , L@S '22: Proceedings of the Ninth ACM Conference on Learning @ Scale)

   Personalized learning stems from the idea that students benefit from instructional material tailored to their needs. Many online learning platforms purport to implement some form of personalized learning, often through on-demand tutoring or self-paced instruction, but to our knowledge none have a way to automatically explore for specific opportunities to personalize students’ education nor a transparent way to identify the effects of personalization on specific groups of students. In this work we present the Automatic Personalized Learning Service (APLS). The APLS uses multi-armed bandit algorithms to recommend the most effective support to each student that requests assistance when completing their online work, and is currently used by ASSISTments, an online learning platform. The first empirical study of the APLS found that Beta-Bernoulli Thompson Sampling, a popular and effective multi-armed bandit algorithm, was only slightly more capable of selecting helpful support than randomly selecting from the relevant support options. Therefore, we also present Decision Tree Thompson Sampling (DTTS), a novel contextual multi-armed bandit algorithm that integrates the transparency and interpretability of decision trees into Thomson sampling. In simulation, DTTS overcame the challenges of recommending support within an online learning platform and was able to increase students’ learning by as much as 10% more than the current algorithm used by the APLS. We demonstrate that DTTS is able to identify qualitative interactions that not only help determine the most effective support for students, but that also generalize well to new students, problems, and support content. The APLS using DTTS is now being deployed at scale within ASSISTments and is a promising tool for all educational learning platforms. 

   more » « less
3. Automatic Interpretable Personalized Learning

   https://doi.org/10.1145/3491140.3528267  

   Prihar, Ethan ; Haim, Aaron ; Sales, Adam ; Heffernan, Neil ( June 2022 , Proceedings of the Ninth ACM Conference on Learning @ Scale (L@S ’22)

   Personalized learning stems from the idea that students benefit from instructional material tailored to their needs. Many online learning platforms purport to implement some form of personalized learning, often through on-demand tutoring or self-paced instruction, but to our knowledge none have a way to automatically explore for specific opportunities to personalize students’ education nor a transparent way to identify the effects of personalization on specific groups of students. In this work we present the Automatic Personalized Learning Service (APLS). The APLS uses multi-armed bandit algorithms to recommend the most effective support to each student that requests assistance when completing their online work, and is currently used by ASSISTments, an online learning platform. The first empirical study of the APLS found that Beta-Bernoulli Thompson Sampling, a popular and effective multi-armed bandit algorithm, was only slightly more capable of selecting helpful support than randomly selecting from the relevant support options. Therefore, we also present Decision Tree Thompson Sampling (DTTS), a novel contextual multi-armed bandit algorithm that integrates the transparency and interpretability of decision trees into Thomson sampling. In simulation, DTTS overcame the challenges of recommending support within an online learning platform and was able to increase students’ learning by as much as 10% more than the current algorithm used by the APLS. We demonstrate that DTTS is able to identify qualitative interactions that not only help determine the most effective support for students, but that also generalize well to new students, problems, and support content. The APLS using DTTS is now being deployed at scale within ASSISTments and is a promising tool for all educational learning platforms. 

   more » « less
4. Thompson-Sampling-Based Wireless Transmission for Panoramic Video Streaming

   Chen, Jiangong ; Li, Bin ; Srikant, R. ( June 2020 , 2020 18th International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOPT))

   Panoramic video streaming has received great attention recently due to its immersive experience. Different from traditional video streaming, it typically consumes 4≈ 6× larger bandwidth with the same resolution. Fortunately, users can only see a portion (roughly 20%) of 360° scenes at each time and thus it is sufficient to deliver such a portion, namely Field of View (FoV), if we can accurately predict user’s motion. In practice, we usually deliver a portion larger than FoV to tolerate inaccurate prediction. Intuitively, the larger the delivered portion, the higher the prediction accuracy. This however leads to a lower transmission success probability. The goal is to select an appropriate delivered portion to maximize system throughput, which can be formulated as a multi-armed bandit problem, where each arm represents the delivered portion. Different from traditional bandit problems with single feedback information, we have two-level feedback information (i.e., both prediction and transmission outcomes) after each decision on the selected portion. As such, we propose a Thompson Sampling algorithm based on two-level feedback information, and demonstrate its superior performance than its traditional counterpart via simulations. 

   more » « less
5. Thompson Sampling for Robust Transfer in Multi-Task Bandits

   Wang, Z. ; Zhang, C. ; Chaudhuri, K. ( January 2022 , Proceedings of the 39th International Conference on Machine Learning)

   We study the problem of online multi-task learning where the tasks are performed within similar but not necessarily identical multi-armed bandit environments. In particular, we study how a learner can improve its overall performance across multiple related tasks through robust transfer of knowledge. While an upper confidence bound (UCB)-based algorithm has recently been shown to achieve nearly-optimal performance guarantees in a setting where all tasks are solved concurrently, it remains unclear whether Thompson sampling (TS) algorithms, which have superior empirical performance in general, share similar theoretical properties. In this work, we present a TS-type algorithm for a more general online multi-task learning protocol, which extends the concurrent setting. We provide its frequentist analysis and prove that it is also nearly-optimal using a novel concentration inequality for multi-task data aggregation at random stopping times. Finally, we evaluate the algorithm on synthetic data and show that the TS-type algorithm enjoys superior empirical performance in comparison with the UCB-based algorithm and a baseline algorithm that performs TS for each individual task without transfer. 

   more » « less