An official website of the United States government
Modern deep learning methods provide effective means to learn good representations. However, is a good representation itself sufficient for sample efficient reinforcement learning? This question has largely been studied only with respect to (worst-case) approximation error, in the more classical approximate dynamic programming literature. With regards to the statistical viewpoint, this question is largely unexplored, and the extant body of literature mainly focuses on conditions which permit sample efficient reinforcement learning with little understanding of what are necessary conditions for efficient reinforcement learning. This work shows that, from the statistical viewpoint, the situation is far subtler than suggested by the more traditional approximation viewpoint, where the requirements on the representation that suffice for sample efficient RL are even more stringent. Our main results provide sharp thresholds for reinforcement learning methods, showing that there are hard limitations on what constitutes good function approximation (in terms of the dimensionality of the representation), where we focus on natural representational conditions relevant to value-based, model-based, and policy-based learning. These lower bounds highlight that having a good (value-based, model-based, or policy-based) representation in and of itself is insufficient for efficient reinforcement learning, unless the quality of this approximation passes certain hard thresholds. Furthermore, our lower bounds also imply exponential separations on the sample complexity between 1) value-based learning with perfect representation and value-based learning with a good-but-not-perfect representation, 2) value-based learning and policy-based learning, 3) policy-based learning and supervised learning and 4) reinforcement learning and imitation learning.
more »
« less
Community Science as Adult Learning: Using Theory to Understand Volunteers’ Experiences
This study explores volunteer learning in an online community science program. Findings indicate alignment with self-directed and experiential learning theory, with implications for learner feedback and engagement.
more »
« less
- Award ID(s):
- 2303019
- PAR ID:
- 10533633
- Publisher / Repository:
- Adult Education Research Conference.
- Date Published:
- Subject(s) / Keyword(s):
- community science, adult STEM education, self-directed learning, experiential learning
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Although considerable research over the past two decades has examined collective learning in environmental governance, much of this scholarship has focused on cases where learning occurred, limiting our understanding of the drivers and barriers to learning. To advance knowledge of what we call the “collective learning continuum,” we compare cases of learning to cases where learning was not found to occur or its effects were “blocked.” Through semi‐structured interviews with key stakeholders in science‐policy forums in the Colorado River Basin, a large and complex river basin in western North America, we examine differences and patterns that explain moments of learning, blocked learning, or non‐learning, drawing insights from the collective learning framework. Our results find various factors that influence learning, blocked learning, and non‐learning. We discover technical and social factors as common drivers of both learning and blocked learning. In contrast, we find more structural factors associated with non‐learning. At the same time, the cases reveal insights about the role of political factors, such as timing, legal constraints, and priorities, which are underdeveloped in the collective learning framework. Overall, these findings advance theoretical knowledge of the collective learning continuum and offer practical insights that may strengthen the coordination of science and management for effective governance within the Basin.more » « less
-
Drawing on parallels with biological immunity, this paper introduces a new use case for learning at the edge called survival-critical machine learning (SCML). Unlike federated learning, which assumes supervised learning with pre-labeled data, SCML involves semi-supervised learning in streaming settings where labels may need to be obtained at very low network bandwidth and extreme class imbalance. We show that the recently-developed workflow of Live Learning is a good fit for SCML. Starting from a weak bootstrap model, this workflow seamlessly pipelines semi-supervised learning, active learning, and transfer learning, with asynchronous bandwidth-sensitive data transmission for labeling. As improved models evolve at the edge through periodic re-training, the threat detection ability of the SCML system improves. This, in turn, improves the survivability of the host system.more » « less
-
null (Ed.)The researchers conducted a qualitative analysis of the perceptions of school personnel and pre-service teachers about an Algebra I tutoring program for students with learning disabilities. The researchers surveyed and interviewed the participants about the effectiveness of the program for the mathematics learning of the students with LD at the school and as a learning experience for the pre-service teachers. The school personnel indicated there was a mutually beneficial relationship between the tutors and the school. The perceptions of the tutors revealed considerations about the challenges they face as Algebra I tutors of students with learning disabilities including remembering Algebra I content, posing strategic questions to students, dealing with students’ math anxiety, and conveying Algebra I content accessibly. The tutors reported positive experiences in the program including learning from field experience and, in particular, learning to promote the progress and learning of students with learning disabilities using gestures and strategic questioning.more » « less
-
This paper unifies commonly used accelerated stochastic gradient methods (Polyak’s Heavy Ball, Nesterov’s Accelerated Gradient and Adaptive Moment Estimation (Adam)) as specific cases of a general lowpass regularized learning framework, the Automatic Stochastic Gradient Method (AutoSGM). For AutoSGM, we derive an optimal iteration-dependent learning rate function and realize an approximation. Adam is also an approximation of this optimal approach that replaces the iteration-dependent learning-rate with a constant. Empirical results on deep neural networks comparing the learning behavior of AutoSGM equipped with this iteration-dependent learning-rate algorithm demonstrate fast learning behavior, robustness to the initial choice of the learning rate, and can tune an initial constant learning-rate in applications where a good constant learning rate approximation is unknown.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
