In this paper, we explore the potential of utilizing time-stamps as labels for Deep Learning from webcams, surveillance cameras, and other fixed viewpoint image situations. Specifically, we explore if learning to classify images by the time they were taken uncovers interesting patterns and behaviors in the scenes captured by these cameras. We describe approaches to building datasets with large quantities of images and their accompanying labels, making them suitable for large-scale deep learning approaches. We share our results from the initial deep learning experiments.
more »
« less
Learning arbitrary stimulus-reward associations for naturalistic stimuli involves transition from learning about features to learning about objects
More Like this
-
-
Undergraduate research is increasingly prevalent in many fields of study, but it is not yet widespread in mathematics education. We argue that expanding undergraduate research opportunities in mathematics education would be beneficial to the field. Such opportunities can be impactful as either extracurricular or course-embedded experiences. To help readers envision directions for undergraduate research experiences in mathematics education with prospective teachers, we describe a model built on a design-based research paradigm. The model engages pairs of prospective teachers in working with faculty mentors to design instructional sequences and test the extent to which they support children’s learning. Undergraduates learn about the nature of systematic mathematics education research and how careful analyses of classroom data can guide practice. Mentors gain opportunities to pursue their personal research interests while guiding undergraduate pairs. We explain how implementing the core cycle of the model, whether on a small or large scale, can help teachers make instructional decisions that are based on rich, qualitative classroom data.more » « less
-
This paper looks at technological advances in the collection and use of information about learning. Updating earlier discussions on transcripts and videorecording, the project used for examples in this paper features a digital simulation and pedagogical patterns where students met through videoconferencing as a post-pandemic alternative to table-base groupwork and then submitted transcripts of the meeting for evaluation and feedback. The transcripts were computationally analyzed to produce data streams showing the shape of conversations. These data were combined with records of students working on collaborative documents and the learning analytics for a digital simulation to illustrate new possibilities to depict collaborative student activity. Congruent with prior reflections on transcription of recordings and the recording process, this paper highlights the ways, old and new, the inscriptional process is not theory-neutral. It privileges certain activities and plays an agentive evidentiary role.more » « less
-
null (Ed.)There is an ongoing debate about whether human rights standards have changed over the last 30 years. The evidence for or against this shift relies upon indicators created by human coders reading the texts of human rights reports. To help resolve this debate, we suggest translating the question of changing standards into a supervised learning problem. From this perspective, the application of consistent standards over time implies a time-constant mapping from the textual features in reports to the human coded scores. Alternatively, if the meaning of abuses have evolved over time, then the same textual features will be labeled with different numerical scores at distinct times. Of course, while the mapping from natural language to numerical human rights score is a highly complicated function, we show that these two distinct data generation processes imply divergent overall patterns of accuracy when we train a wide variety of algorithms on older versus newer sets of observations to learn how to automatically label texts with scores. Our results are consistent with the expectation that standards of human rights have changed over time.more » « less
-
more » « less
Summary Many studies have demonstrated that illustrating expository science texts with images that are interesting, but irrelevant for understanding the causal relations underlying scientific phenomena, can cause seduction effects, which can reduce understanding from text. The term “seduction effects” refers to the influence that images are thought to have on readers, seducing them away from deeply processing important information. The present study explores whether images relevant for instructional goals may also show some seduction effects. In this study, the presence of photographic images negatively impacted understanding compared with the presence of relevant animations or instructing students to sketch a drawing during reading. However, the results showed that both photographic images and relevant animations could lead to illusions of understanding, whereas sketching did not. The results suggest that even images that are relevant for instructional goals may sometimes result in seduction effects that deceive readers when judging their own understanding.
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1016/j.cognition.2020.104425
