Matching Items (3)
Filtering by
- All Subjects: Transistors
- All Subjects: Feature Learning
- Creators: Thornton, Trevor
Description
Feature representations for raw data is one of the most important component in a machine learning system. Traditionally, features are \textit{hand crafted} by domain experts which can often be a time consuming process. Furthermore, they do not generalize well to unseen data and novel tasks. Recently, there have been many efforts to generate data-driven representations using clustering and sparse models. This dissertation focuses on building data-driven unsupervised models for analyzing raw data and developing efficient feature representations.
Simultaneous segmentation and feature extraction approaches for silicon-pores sensor data are considered. Aggregating data into a matrix and performing low rank and sparse matrix decompositions with additional smoothness constraints are proposed to solve this problem. Comparison of several variants of the approaches and results for signal de-noising and translocation/trapping event extraction are presented. Algorithms to improve transform-domain features for ion-channel time-series signals based on matrix completion are presented. The improved features achieve better performance in classification tasks and in reducing the false alarm rates when applied to analyte detection.
Developing representations for multimedia is an important and challenging problem with applications ranging from scene recognition, multi-media retrieval and personal life-logging systems to field robot navigation. In this dissertation, we present a new framework for feature extraction for challenging natural environment sounds. Proposed features outperform traditional spectral features on challenging environmental sound datasets. Several algorithms are proposed that perform supervised tasks such as recognition and tag annotation. Ensemble methods are proposed to improve the tag annotation process.
To facilitate the use of large datasets, fast implementations are developed for sparse coding, the key component in our algorithms. Several strategies to speed-up Orthogonal Matching Pursuit algorithm using CUDA kernel on a GPU are proposed. Implementations are also developed for a large scale image retrieval system. Image-based "exact search" and "visually similar search" using the image patch sparse codes are performed. Results demonstrate large speed-up over CPU implementations and good retrieval performance is also achieved.
Simultaneous segmentation and feature extraction approaches for silicon-pores sensor data are considered. Aggregating data into a matrix and performing low rank and sparse matrix decompositions with additional smoothness constraints are proposed to solve this problem. Comparison of several variants of the approaches and results for signal de-noising and translocation/trapping event extraction are presented. Algorithms to improve transform-domain features for ion-channel time-series signals based on matrix completion are presented. The improved features achieve better performance in classification tasks and in reducing the false alarm rates when applied to analyte detection.
Developing representations for multimedia is an important and challenging problem with applications ranging from scene recognition, multi-media retrieval and personal life-logging systems to field robot navigation. In this dissertation, we present a new framework for feature extraction for challenging natural environment sounds. Proposed features outperform traditional spectral features on challenging environmental sound datasets. Several algorithms are proposed that perform supervised tasks such as recognition and tag annotation. Ensemble methods are proposed to improve the tag annotation process.
To facilitate the use of large datasets, fast implementations are developed for sparse coding, the key component in our algorithms. Several strategies to speed-up Orthogonal Matching Pursuit algorithm using CUDA kernel on a GPU are proposed. Implementations are also developed for a large scale image retrieval system. Image-based "exact search" and "visually similar search" using the image patch sparse codes are performed. Results demonstrate large speed-up over CPU implementations and good retrieval performance is also achieved.
ContributorsSattigeri, Prasanna S (Author) / Spanias, Andreas (Thesis advisor) / Thornton, Trevor (Committee member) / Goryll, Michael (Committee member) / Tsakalis, Konstantinos (Committee member) / Arizona State University (Publisher)
Created2014
Description
The Metal Semiconductor Field Effect Transistor (MESFET) has high potential to enter analog and RF applications due to their high breakdown voltage and switching frequency characteristics. These MESFET devices could allow for high voltage analog circuits to be integrated with low voltage digital circuits on a single chip in an extremely cost effective way. Higher integration leads to electronics with increased functionality and a smaller finished product. The MESFETs are designed in-house by the research group led by Dr. Trevor Thornton. The layouts are then sent to multi-project wafer (MPW) integrated circuit foundry companies, such as the Metal Oxide Semiconductor Implementation Service (MOSIS) to be fabricated. Once returned, the electrical characteristics of the devices are measured. The MESFET has been implemented in various applications by the research group, including the low dropout linear regulator (LDO) and RF power amplifier. An advantage of the MESFET is that it can function in extreme environments such as space, allowing for complex electrical systems to continue functioning properly where traditional transistors would fail.
ContributorsKam, Jason (Author) / Thornton, Trevor (Thesis director) / Goryll, Michael (Committee member) / Barrett, The Honors College (Contributor) / Electrical Engineering Program (Contributor)
Created2015-05
Description
In this work, the insight provided by our sophisticated Full Band Monte Carlo simulator is used to analyze the behavior of state-of-art devices like GaN High Electron Mobility Transistors and Hot Electron Transistors. Chapter 1 is dedicated to the description of the simulation tool used to obtain the results shown in this work. Moreover, a separate section is dedicated the set up of a procedure to validate to the tunneling algorithm recently implemented in the simulator. Chapter 2 introduces High Electron Mobility Transistors (HEMTs), state-of-art devices characterized by highly non linear transport phenomena that require the use of advanced simulation methods. The techniques for device modeling are described applied to a recent GaN-HEMT, and they are validated with experimental measurements. The main techniques characterization techniques are also described, including the original contribution provided by this work. Chapter 3 focuses on a popular technique to enhance HEMTs performance: the down-scaling of the device dimensions. In particular, this chapter is dedicated to lateral scaling and the calculation of a limiting cutoff frequency for a device of vanishing length. Finally, Chapter 4 and Chapter 5 describe the modeling of Hot Electron Transistors (HETs). The simulation approach is validated by matching the current characteristics with the experimental one before variations of the layouts are proposed to increase the current gain to values suitable for amplification. The frequency response of these layouts is calculated, and modeled by a small signal circuit. For this purpose, a method to directly calculate the capacitance is developed which provides a graphical picture of the capacitative phenomena that limit the frequency response in devices. In Chapter 5 the properties of the hot electrons are investigated for different injection energies, which are obtained by changing the layout of the emitter barrier. Moreover, the large signal characterization of the HET is shown for different layouts, where the collector barrier was scaled.
ContributorsSoligo, Riccardo (Author) / Saraniti, Marco (Thesis advisor) / Goodnick, Stephen M (Committee member) / Chowdhury, Srabanti (Committee member) / Thornton, Trevor (Committee member) / Arizona State University (Publisher)
Created2016