Search

Active learning aims to reduce the time and cost of developing speech recognition systems by selecting for transcription highly informative subsets from large pools of audio data. Previous evaluations at OpenKWS and IARPA BABEL have investigated data selection for low-resource languages in very constrained scenarios with 2-hour data selections given a 1-hour seed set. We expand on this to investigate what happens with larger selections and fewer constraints on language modeling data. Our results, on four languages from the final BABEL OP3 period, show that active learning is helpful at larger selections with consistent gains up to 14 hours. We also find that the impact of additional language model data is orthogonal to the impact of the active learning selection criteria.

Automatic speech recognition (ASR) systems rely on large quantities of transcribed acoustic data. The collection of audio data is relatively cheap, whereas the transcription of that data is relatively expensive. Thus there is an interest in the ASR community in active learning, in which only a small subset of highly representative data chosen from a large pool of untranscribed audio need be transcribed in order to approach the performance of the system trained with much larger amounts of transcribed audio. In this paper, we compare two basic approaches to active learning: a supervised approach in which we build a speech recognition system from a small amount of seed data in order to make the selection of a limited amount of additional audio for transcription, and an unsupervised approach in which no intermediate system recognition system built from seed data is necessary. Our best unsupervised approach performs quite close to our supervised approach, with both outperforming a random selection scheme.

We demonstrate the use of sequence pattern mining as applied to monitoring the usage of emailing software by clients with cognitive impairments. We show how Max Motif, a sequence-mining algorithm, can be applied using a stream-mining approach. Consequently, clinicians can now consider sequential patterns in their analysis of client emailing behaviors. Such analysis is part of the Think and Link project, which provides personalized email clients, a kind of assistive technology, to aid client communications that facilitate activities of daily living. By using the simplified, customized system, clients can now email, whereas they could not previously. By continuously monitoring usage, the project is able continually adapt the software to a user's changing needs. Thus, monitoring software usage, particularly email event sequences, is important. This paper introduces theory and design of stream sequence-mining for UI event streams.

Personalized requirements practices can be applied to specify goals as part of a clinical plan to aid cognitive rehabilitation. In this context, requirements monitoring can aid clinicians in tracking user behaviors as they attempt to achieve their goals. Quality metrics over stream-mined models can identify potential changes in user goal attainment, as a user learns his or her personalized emailing system. When the quality of some models varies significantly from nearby models-as defined by quality metrics-then the user's behavior is automatically flagged as a potentially significant behavioral change. The specific changes in user behavior can be automatically derived by differencing the data-mined decision-tree model. This paper describes how decision tree differencing can aid diagnoses of behavioral changes in a case study of cognitive rehabilitation via emailing. The technique may be more widely applicable to other requirements monitoring contexts.