Matching Items (8)
Filtering by
- Genre: Masters Thesis
- Creators: Arizona State University
- Creators: Chakraborty, Bijeta
- Member of: ASU Electronic Theses and Dissertations
Description
We solve the problem of activity verification in the context of sustainability. Activity verification is the process of proving the user assertions pertaining to a certain activity performed by the user. Our motivation lies in incentivizing the user for engaging in sustainable activities like taking public transport or recycling. Such incentivization schemes require the system to verify the claim made by the user. The system verifies these claims by analyzing the supporting evidence captured by the user while performing the activity. The proliferation of portable smart-phones in the past few years has provided us with a ubiquitous and relatively cheap platform, having multiple sensors like accelerometer, gyroscope, microphone etc. to capture this evidence data in-situ. In this research, we investigate the supervised and semi-supervised learning techniques for activity verification. Both these techniques make use the data set constructed using the evidence submitted by the user. Supervised learning makes use of annotated evidence data to build a function to predict the class labels of the unlabeled data points. The evidence data captured can be either unimodal or multimodal in nature. We use the accelerometer data as evidence for transportation mode verification and image data as evidence for recycling verification. After training the system, we achieve maximum accuracy of 94% when classifying the transport mode and 81% when detecting recycle activity. In the case of recycle verification, we could improve the classification accuracy by asking the user for more evidence. We present some techniques to ask the user for the next best piece of evidence that maximizes the probability of classification. Using these techniques for detecting recycle activity, the accuracy increases to 93%. The major disadvantage of using supervised models is that it requires extensive annotated training data, which expensive to collect. Due to the limited training data, we look at the graph based inductive semi-supervised learning methods to propagate the labels among the unlabeled samples. In the semi-supervised approach, we represent each instance in the data set as a node in the graph. Since it is a complete graph, edges interconnect these nodes, with each edge having some weight representing the similarity between the points. We propagate the labels in this graph, based on the proximity of the data points to the labeled nodes. We estimate the performance of these algorithms by measuring how close the probability distribution of the data after label propagation is to the probability distribution of the ground truth data. Since labeling has a cost associated with it, in this thesis we propose two algorithms that help us in selecting minimum number of labeled points to propagate the labels accurately. Our proposed algorithm achieves a maximum of 73% increase in performance when compared to the baseline algorithm.
ContributorsDesai, Vaishnav (Author) / Sundaram, Hari (Thesis advisor) / Li, Baoxin (Thesis advisor) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2013
Description
Efficiency of components is an ever increasing area of importance to portable applications, where a finite battery means finite operating time. Higher efficiency devices need to be designed that don't compromise on the performance that the consumer has come to expect. Class D amplifiers deliver on the goal of increased efficiency, but at the cost of distortion. Class AB amplifiers have low efficiency, but high linearity. By modulating the supply voltage of a Class AB amplifier to make a Class H amplifier, the efficiency can increase while still maintaining the Class AB level of linearity. A 92dB Power Supply Rejection Ratio (PSRR) Class AB amplifier and a Class H amplifier were designed in a 0.24um process for portable audio applications. Using a multiphase buck converter increased the efficiency of the Class H amplifier while still maintaining a fast response time to respond to audio frequencies. The Class H amplifier had an efficiency above the Class AB amplifier by 5-7% from 5-30mW of output power without affecting the total harmonic distortion (THD) at the design specifications. The Class H amplifier design met all design specifications and showed performance comparable to the designed Class AB amplifier across 1kHz-20kHz and 0.01mW-30mW. The Class H design was able to output 30mW into 16Ohms without any increase in THD. This design shows that Class H amplifiers merit more research into their potential for increasing efficiency of audio amplifiers and that even simple designs can give significant increases in efficiency without compromising linearity.
ContributorsPeterson, Cory (Author) / Bakkaloglu, Bertan (Thesis advisor) / Barnaby, Hugh (Committee member) / Kiaei, Sayfe (Committee member) / Arizona State University (Publisher)
Created2013
Description
Class D Amplifiers are widely used in portable systems such as mobile phones to achieve high efficiency. The demands of portable electronics for low power consumption to extend battery life and reduce heat dissipation mandate efficient, high-performance audio amplifiers. The high efficiency of Class D amplifiers (CDAs) makes them particularly attractive for portable applications. The Digital class D amplifier is an interesting solution to increase the efficiency of embedded systems. However, this solution is not good enough in terms of PWM stage linearity and power supply rejection. An efficient control is needed to correct the error sources in order to get a high fidelity sound quality in the whole audio range of frequencies. A fundamental analysis on various error sources due to non idealities in the power stage have been discussed here with key focus on Power supply perturbations driving the Power stage of a Class D Audio Amplifier. Two types of closed loop Digital Class D architecture for PSRR improvement have been proposed and modeled. Double sided uniform sampling modulation has been used. One of the architecture uses feedback around the power stage and the second architecture uses feedback into digital domain. Simulation & experimental results confirm that the closed loop PSRR & PS-IMD improve by around 30-40 dB and 25 dB respectively.
ContributorsChakraborty, Bijeta (Author) / Bakkaloglu, Bertan (Thesis advisor) / Garrity, Douglas (Committee member) / Ozev, Sule (Committee member) / Arizona State University (Publisher)
Created2012
Description
In this thesis, a digital input class D audio amplifier system which has the ability
to reject the power supply noise and nonlinearly of the output stage is presented. The main digital class D feed-forward path is using the fully-digital sigma-delta PWM open loop topology. Feedback loop is used to suppress the power supply noise and harmonic distortions. The design is using global foundry 0.18um technology.
Based on simulation, the power supply rejection at 200Hz is about -49dB with
81dB dynamic range and -70dB THD+N. The full scale output power can reach as high as 27mW and still keep minimum -68dB THD+N. The system efficiency at full scale is about 82%.
to reject the power supply noise and nonlinearly of the output stage is presented. The main digital class D feed-forward path is using the fully-digital sigma-delta PWM open loop topology. Feedback loop is used to suppress the power supply noise and harmonic distortions. The design is using global foundry 0.18um technology.
Based on simulation, the power supply rejection at 200Hz is about -49dB with
81dB dynamic range and -70dB THD+N. The full scale output power can reach as high as 27mW and still keep minimum -68dB THD+N. The system efficiency at full scale is about 82%.
ContributorsBai, Jing (Author) / Bakkaloglu, Bertan (Thesis advisor) / Arizona State University (Publisher)
Created2015
Description
Machine learning methodologies are widely used in almost all aspects of software engineering. An effective machine learning model requires large amounts of data to achieve high accuracy. The data used for classification is mostly labeled, which is difficult to obtain. The dataset requires both high costs and effort to accurately label the data into different classes. With abundance of data, it becomes necessary that all the data should be labeled for its proper utilization and this work focuses on reducing the labeling effort for large dataset. The thesis presents a comparison of different classifiers performance to test if small set of labeled data can be utilized to build accurate models for high prediction rate. The use of small dataset for classification is then extended to active machine learning methodology where, first a one class classifier will predict the outliers in the data and then the outlier samples are added to a training set for support vector machine classifier for labeling the unlabeled data. The labeling of dataset can be scaled up to avoid manual labeling and building more robust machine learning methodologies.
ContributorsBatra, Salil (Author) / Femiani, John (Thesis advisor) / Amresh, Ashish (Thesis advisor) / Bansal, Ajay (Committee member) / Arizona State University (Publisher)
Created2017
Description
Clickers are a common part of many classrooms across universities. Despite the widespread use, education researchers disagree about how to best use these tools and about how they impact students. Prior work has shown possible differential impacts of clickers based on demographic indicators, such as age, gender, and ethnicity. To explore these topics a two-part project was designed. First, a literature review was completed focusing on past and current clicker practices and the research surrounding them. Second, original data, stratified by demographic characteristics, was collected on student perceptions of clickers. The literature review revealed that not all uses of clickers are created equal. Instructors in higher education first introduced clickers to enhance traditional pedagogies by simplifying common classroom tasks (e.g. grading, attendance, feedback collection). More recently, instructors pair clickers and novel pedagogies. A review of the identified benefits and drawbacks for students and instructors is provided for both approaches. Instructors can use different combinations of technological competency and pedagogical content knowledge that lead to four main outcomes. When instructors have both technological competency and pedagogical content knowledge, all the involved parties, students and instructors, benefit. When instructors have technological competency but lack pedagogical content knowledge, instructors are the main benefactors. When instructors have pedagogical content knowledge alone, students can benefit, but usefulness to the instructor decreases. When instructors have neither technological competency nor pedagogical content knowledge, no party benefits. Beyond these findings, recommendations are provided for future clicker research. Second, the review highlighted that clickers may have a differential impact on students of different demographic groups. To explore this dynamic, an original study on student views of clickers, which included demographic data, was conducted. The original study does not find significantly different enthusiasm for clickers by demographic group, unlike prior studies that explored some of these relationships. However, white students and male students are overrepresented in the group that does not enjoy clickers. This conclusion is supported by visual observations from the means of the demographic groups. Overall, based on the review of the literature and original research, if instructors pair clickers with validated pedagogies, and if researchers continue to study clicker classrooms, including which students like and benefit from clickers, clickers may continue to be a valuable educational technology.
ContributorsChambers, Elijah Lorenzo (Author) / Henderson, Joesph (Thesis advisor) / Ellison, Karin (Thesis advisor) / Chew, Matthew (Committee member) / Arizona State University (Publisher)
Created2020
Description
Science education faces a distinct challenge in the transition to active learning: how can teachers ensure students reach accurate understandings during the exploration and self-discovery phase of a lesson? Research in hypothesis generation demonstrates human's vulnerabilities to specific biases based on prior knowledge, selective memory retrieval, and failure to consider alternative explanations. This is further complicated in science education, where content standards are abstract. As such, it is imperative to implement a proactive intervention to curb misconceptions from forming during active learning in science lessons. In this work, a new a model of instruction, Question-Based Learning (QBL) is designed and tested against current learning paradigms. The study aims to investigate whether providing constraint-seeking questions is an effective intervention leading to improved mastery of learning targets during active learning. Participants were randomly assigned to one of three conditions to learn a scientific concept: a blended learning condition, a guided-inquiry condition, or a QBL condition. Mastery was measured at the end of the task using a 12-question assessment. The same measure was also administered one week after subjects completed the study to see whether delayed recall significantly differs between condition groups. Results indicate the QBL model is at least as effective two existing forms of pedagogy at teaching a scientific principle, increasing depth of knowledge regarding that scientific principle, and sustaining knowledge over time.
ContributorsWallace, Grace Kathleen (Author) / Duran, Nicholas (Thesis advisor) / Lucca, Kelsey (Committee member) / Horne, Zachary (Committee member) / Arizona State University (Publisher)
Created2021
Description
Increasingly, college courses have transitioned from traditional lecture to student-centered active learning, creating more opportunities for students to interact with each other in class. Recent studies have indicated that these increased interactions in active learning can create situations where students’ identities are more salient, which could result in novel challenges for students with marginalized identities. Christianity has been shown to be a marginalized identity in the context of undergraduate biology courses, but it is unknown whether Christian students experience challenges in their interactions with other students in class. The social psychology framework of concealable stigmatized identity (CSI) was used to explore the experiences of Christian students during peer interactions in undergraduate biology courses. Thirty students were interviewed, and most felt their religious identity was salient during peer interactions in biology. Students also reported that they have more opportunities to reveal their religious identity in courses that incorporate peer discussion than in courses that do not. Students claimed that revealing their religious identity to their peers could be beneficial because they could find other religious students in their courses, grow closer with their peers, and combat stereotypes about religious individuals in science. Though most students anticipated stigma, which caused some students to choose not to reveal their religious identities, comparatively few had experienced stigma during peer interactions in their college biology courses, and even fewer had experienced stigma from peers who knew they were religious. These findings indicate that it be may important to teach students how to be culturally competent to reduce Christian students’ anticipated and experienced stigma in active learning courses.
ContributorsEdwards, Baylee Anne (Author) / Brownell, Sara E. (Thesis advisor) / Barnes, M. Elizabeth (Committee member) / Sterner, Beckett (Committee member) / Cooper, Katelyn M. (Committee member) / Arizona State University (Publisher)
Created2022