We study the problem of active learning in a stream-based setting, allowing the distribution of the examples to change over time. We prove upper bounds on the number of prediction mistakes and number of label requests for established disagreement-based active learning algorithms, both in the realizable case and under Tsybakov noise. We further prove minimax lower bounds for this problem.

Active learning is a useful technique for tasks for which unlabeled data is abundant but manual labeling is expensive. One example of such a task is semantic role labeling (SRL), which relies heavily on labels from trained linguistic experts. One challenge in applying active learning algorithms for SRL is that the complete knowledge of the SRL model is often unavailable, against the common assu...

Institutional pressures to make optimal use of lecture halls and classrooms can be powerful motivators to identify resources to develop technology enhanced learning approaches to traditional curricula. From the academic’s perspective, engaging students in active learning and reducing the academic workload are important and complementary drivers. This paper presents a case study of a curriculum ...

Generic ensemble methods can achieve excellent learning performance, but are not good candidates for active learning because of their different design purposes. We investigate how to use diversity of the member classifiers of an ensemble for efficient active learning. We empirically show, using benchmark data sets, that (1) to achieve a good (stable) ensemble, the number of classifiers needed i...

While Active Learning (AL) has already been shown to markedly reduce the annotation efforts for many sequence labeling tasks compared to random selection, AL remains unconcerned about the internal structure of the selected sequences (typically, sentences). We propose a semisupervised AL approach for sequence labeling where only highly uncertain subsequences are presented to human annotators, wh...

introduction: aphasia is a language disorder resulting from a lesion in the cerebral cortex. in this case report, we present a polyglot patient who recovered from aphasia by speaking his newly active learned language case report: a 69 years old male referred with acute onset right hemiparesis and global aphasia. after imaging, he treated with 75 mg r-tpa (0.9 mg/kg). after the fourth day of hos...

In many industrial applications, obtaining labeled observations is not straightforward as it often requires the intervention of human experts or use expensive testing equipment. these circumstances, active learning can be highly beneficial in suggesting most informative data points to used when fitting a model. Reducing number needed for model development alleviates both computational burden re...

Most classic machine learning methods depend on the assumption that humans can annotate all the data available for training. However, many modern machine learning applications (including image and video classification, protein sequence classification, and speech processing) have massive amounts of unannotated or unlabeled data. As a consequence, there has been tremendous interest both in machin...