Search

Recent advances in machine learning have instigated a renewed interest in using machine learning approaches to better understand human sensory processing. This line of research is particularly interesting for speech research since speech comprehension is uniquely human, which complicates obtaining detailed neural recordings. In this thesis, I explore how machine learning can be used to uncover new knowledge about the auditory system, with a focus on discovering robust auditory features. The resulting increased understanding of the noise robustness of human hearing may help to better assist those with hearing loss and improve Automatic Speech Recognition (ASR) systems. First, I show how computational neuroscience and machine learning can be combined to generate hypotheses about auditory features. I introduce a neural feature detection model with a modest number of parameters that is compatible with auditory physiology. By testing feature detector variants in a speech classification task, I confirm the importance of both well-studied and lesser-known auditory features. Second, I investigate whether ASR software is a good candidate model of the human auditory system. By comparing several state-of-the-art ASR systems to the results from humans on a range of psychometric experiments, I show that these ASR systems diverge markedly from humans in at least some psychometric tests. This implies that none of these systems act as a strong proxy for human speech recognition, although some may be useful when asking more narrowly defined questions. For neuroscientists, this thesis exemplifies how machine learning can be used to generate new hypotheses about human hearing, while also highlighting the caveats of investigating systems that may work fundamentally differently from the human brain. For machine learning engineers, I point to tangible directions for improving ASR systems. To motivate the continued cross-fertilization between these fields, a toolbox that allows researchers to assess new ASR systems has been released.

The increase in numbers of students with disabilities in American school districts has created the need to study options that have proven successful in meeting a variety of students’ educational achievement needs in regular education classrooms. The use of co-teaching as a methodology to help meet this need is a worthy research topic. A mixed methods study was conducted seeking information from special education and regular education middle school teachers in a select Minnesota school regarding co-teaching effectiveness, challenges, and benefits to students. The study participants were selected because they utilized co-teaching methodologies.  The study may provide guidelines for teachers currently employing one of the six researched co-teaching models and for administrators contemplating the use of co-teaching models in the classroom in the future. Statement of the Problem Co-teaching methodology pairs a special education and general education teachers in the same classroom; this model was initially developed to provide instruction to students with disabilities in the regular classroom for a portion of their day rather than serving these students extensively in a resource room (Friend & Cook, 1995).  Between 2011-12 and 2017-18, students in public schools who received special education services increased from 6.4 million to 7 million (U.S. Department of Education, 2019). That increase in the number of students with disabilities created a need to study options that have proven successful in meeting a variety of students’ educational achievement needs in regular education classrooms. Study Purpose and Overview The purpose of the mixed methods study was to examine the effectiveness and challenges of co-teaching, the benefits to students, and the most frequently employed co-teaching model from the Friend and Cook research, 2010, as reported by special education and general education teachers in select Minnesota middle schools who have implemented a co-teaching model in their classrooms. The study also examined the perceived differences in challenges identified by each co-teaching member.   Key Findings Significant findings from the survey included: 62.5% of study participants perceived the Parallel Teaching model as very effective. 93.8 % of study participants either agreed or strongly agreed that co-teaching had increased the academic achievement of special education students. 93.8 % of study participants either agreed or strongly agreed that co-teaching allowed for increased opportunities to provide differentiated instruction. 87.6 % of study participants either agreed or strongly agreed that a lack of planning time between co-teaching partners was a challenge. The co-teaching model allowed teachers to differentiate instruction and reflect on what was beneficial to student achievement and to quality teaching. Implementation of a co-teaching model requires advanced planning, training of teachers and scheduled time to meet during the school day and week. Challenges occur when placing two teachers’ personalities together. It is essential to ensure that the teachers will be able to work together and truly want to be part of the co-teaching team.