Matching Items (120)
Filtering by
- Creators: Yang, Yezhou
- Member of: ASU Electronic Theses and Dissertations
- Status: Published
Description
Small molecules have proven to be very important tools for exploration of biological systems including diagnosis and treatment of lethal diseases like cancer. Fluorescent probes have been extensively used to further amplify the utilization of small molecules. The manipulation of naturally occurring biological targets with the help of synthetic compounds is the focus of the work described in this thesis.
Bleomycins (BLMs) are a class of water soluble, glycopeptide-derived antitumor antibiotics consisting of a structurally complicated unnatural hexapeptide and a disaccharide, clinically used as an anticancer chemotherapeutic agent at an exceptionally low therapeutic dose. The efficiency of BLM is likely achieved both by selective localization within tumor cells and selective binding to DNA followed by efficient double-strand cleavage. The disaccharide moiety is responsible for the tumor cell targeting properties of BLM. A recent study showed that both BLM and its disaccharide, conjugated to the cyanine dye Cy5**, bound selectively to cancer cells. Thus, the disaccharide moiety alone recapitulates the tumor cell targeting properties of BLM. Work presented here describes the synthesis of the fluorescent carbohydrate conjugates. A number of dye-labeled modified disaccharides and monosaccharides were synthesized to study the nature of the participation of the carbamoyl moiety in the mechanism of tumor cell recognition and uptake by BLM saccharides. It was demonstrated that the carbamoylmannose moiety of BLM is the smallest structural entity capable for the cellular targeting and internalization, and the carbamoyl functionality is indispensible for tumor cell targeting. It was also confirmed that BLM is a modular molecule, composed of a tumor cell targeting moiety (the saccharide) attached to a cytotoxic DNA cleaving domain (the BLM aglycone). These finding encouraged us to further synthesize carbohydrate probes for PET imaging and to conjugate the saccharide moiety with cytotoxins for targeted delivery to tumor cells.
The misacylated suppressor tRNA technique has enabled the site-specific incorporation of noncanonical amino acids into proteins. The focus of the present work was the synthesis of unnatural lysine analogues with nucleophilic properties for incorporation at position 72 of the lyase domain of human DNA polymerase beta, a multifunctional enzyme with dRP lyase and polymerase activity.
Bleomycins (BLMs) are a class of water soluble, glycopeptide-derived antitumor antibiotics consisting of a structurally complicated unnatural hexapeptide and a disaccharide, clinically used as an anticancer chemotherapeutic agent at an exceptionally low therapeutic dose. The efficiency of BLM is likely achieved both by selective localization within tumor cells and selective binding to DNA followed by efficient double-strand cleavage. The disaccharide moiety is responsible for the tumor cell targeting properties of BLM. A recent study showed that both BLM and its disaccharide, conjugated to the cyanine dye Cy5**, bound selectively to cancer cells. Thus, the disaccharide moiety alone recapitulates the tumor cell targeting properties of BLM. Work presented here describes the synthesis of the fluorescent carbohydrate conjugates. A number of dye-labeled modified disaccharides and monosaccharides were synthesized to study the nature of the participation of the carbamoyl moiety in the mechanism of tumor cell recognition and uptake by BLM saccharides. It was demonstrated that the carbamoylmannose moiety of BLM is the smallest structural entity capable for the cellular targeting and internalization, and the carbamoyl functionality is indispensible for tumor cell targeting. It was also confirmed that BLM is a modular molecule, composed of a tumor cell targeting moiety (the saccharide) attached to a cytotoxic DNA cleaving domain (the BLM aglycone). These finding encouraged us to further synthesize carbohydrate probes for PET imaging and to conjugate the saccharide moiety with cytotoxins for targeted delivery to tumor cells.
The misacylated suppressor tRNA technique has enabled the site-specific incorporation of noncanonical amino acids into proteins. The focus of the present work was the synthesis of unnatural lysine analogues with nucleophilic properties for incorporation at position 72 of the lyase domain of human DNA polymerase beta, a multifunctional enzyme with dRP lyase and polymerase activity.
ContributorsBhattacharya, Chandrabali (Author) / Hecht, Sidney M. (Thesis advisor) / Moore, Ana (Committee member) / Gould, Ian R (Committee member) / Arizona State University (Publisher)
Created2014
Description
The hydrothermal chemistry of organic compounds influences many critical geological processes, including the formation of oil and gas reservoirs, the degradation and transport of organic matter in sedimentary basins, metabolic cycles in the deep subsurface biosphere, and possibly prebiotic organic synthesis related to the origin of life. In most previous studies of hydrothermal organic reactions the emphasis has been mainly on determining reaction product distributions, studies that provide detailed mechanistic information or direct evidence for specific reaction intermediates are rare. To develop a better understanding, I performed hydrothermal experiments with model ketone compound dibenzylketone (DBK), which serves as a quite useful tool to probe the bond breaking and forming processes in hydrothermal geochemical transformations. A careful study of reaction kinetics and products of DBK in Chapter 2 of this dissertation reveals reversible and irreversible reaction pathways, and provides evidence for competing ionic and radical reaction mechanisms. The majority of the observed products result from homolytic carbon-carbon and carbon-hydrogen bond cleavage and secondary coupling reactions of the benzyl and related radical intermediates.
In the third chapter of the dissertation, a novel hydrothermal photochemical method is studied, which enabled in situ independent generation of the relevant radicals and effectively separated the radical and ionic reactions that occur simultaneously in pure thermal reactions. In the following chapter, I focus on the role of minerals on ketone hydrothermal reactions. Minerals such as quartz and corundum have no detectable effect on DBK, whereas magnetite, hematite, and troilite all increase ketone reactivity to various extents. The influence of these iron-bearing minerals can be attributed to the mineral surface catalysis or the solution chemistry change that is presumably caused by dissolved inorganic species from minerals. In addition, some new discoveries on strong oxidizing effect of copper (II) ion under hydrothermal conditions are described in the latter chapter of the dissertation, where examples of clean and rapid reactions that converted alcohols to aldehyde and aldehydes to carboxylic acids are included.
In the third chapter of the dissertation, a novel hydrothermal photochemical method is studied, which enabled in situ independent generation of the relevant radicals and effectively separated the radical and ionic reactions that occur simultaneously in pure thermal reactions. In the following chapter, I focus on the role of minerals on ketone hydrothermal reactions. Minerals such as quartz and corundum have no detectable effect on DBK, whereas magnetite, hematite, and troilite all increase ketone reactivity to various extents. The influence of these iron-bearing minerals can be attributed to the mineral surface catalysis or the solution chemistry change that is presumably caused by dissolved inorganic species from minerals. In addition, some new discoveries on strong oxidizing effect of copper (II) ion under hydrothermal conditions are described in the latter chapter of the dissertation, where examples of clean and rapid reactions that converted alcohols to aldehyde and aldehydes to carboxylic acids are included.
ContributorsYang, Ziming (Author) / Shock, Everett L (Thesis advisor) / Gould, Ian R (Committee member) / Wolf, George H. (Committee member) / Arizona State University (Publisher)
Created2014
Description
Mitochondria are crucial intracellular organelles which play a pivotal role in providing energy to living organisms in the form of adenosine triphosphate (ATP). The mitochondrial electron transport chain (ETC) coupled with oxidative phosphorylation (OX-PHOS) transforms the chemical energy of amino acids, fatty acids and sugars to ATP. The mitochondrial electron transport system consumes nearly 90% of the oxygen used by the cell. Reactive oxygen species (ROS) in the form of superoxide anions (O2*-) are generated as byproduct of cellular metabolism due to leakage of electrons from complex I and complex III to oxygen. Under normal conditions, the effects of ROS are offset by a variety of antioxidants (enzymatic and non-enzymatic).
Mitochondrial dysfunction has been proposed in the etiology of various pathologies, including cardiovascular and neurodegenerative diseases, such as Alzheimer's disease and Parkinson's disease, ischemia-reperfusion (IR) injury, diabetes and aging. To treat these disorders, it is imperative to target mitochondria, especially the electron transport chain. One of the methodologies currently used for the treatment of mitochondrial and neurodegenerative diseases where endogenous antioxidant defenses are inadequate for protecting against ROS involves the administration of exogenous antioxidants.
As part of our pursuit of effective neuroprotective drugs, a series of pyridinol and pyrimidinol analogues have been rationally designed and synthesized. All the analogues were evaluated for their ability to quench lipid peroxidation and reactive oxygen species (ROS), and preserve mitochondrial membrane potential (Δm) and support ATP synthesis. These studies are summarized in Chapter 2.
Drug discovery and lead identification can be reinforced by assessing the metabolic fate of orally administered drugs using simple microsomal incubation experiments. Accordingly, in vitro microsomal studies were designed and carried out using bovine liver microsomes to screen available pyridinol and pyrimidinol analogues for their metabolic lability. The data obtained was utilized for an initial assessment of potential bioavailability of the compounds screened and is summarized fully in Chapter 3.
Mitochondrial dysfunction has been proposed in the etiology of various pathologies, including cardiovascular and neurodegenerative diseases, such as Alzheimer's disease and Parkinson's disease, ischemia-reperfusion (IR) injury, diabetes and aging. To treat these disorders, it is imperative to target mitochondria, especially the electron transport chain. One of the methodologies currently used for the treatment of mitochondrial and neurodegenerative diseases where endogenous antioxidant defenses are inadequate for protecting against ROS involves the administration of exogenous antioxidants.
As part of our pursuit of effective neuroprotective drugs, a series of pyridinol and pyrimidinol analogues have been rationally designed and synthesized. All the analogues were evaluated for their ability to quench lipid peroxidation and reactive oxygen species (ROS), and preserve mitochondrial membrane potential (Δm) and support ATP synthesis. These studies are summarized in Chapter 2.
Drug discovery and lead identification can be reinforced by assessing the metabolic fate of orally administered drugs using simple microsomal incubation experiments. Accordingly, in vitro microsomal studies were designed and carried out using bovine liver microsomes to screen available pyridinol and pyrimidinol analogues for their metabolic lability. The data obtained was utilized for an initial assessment of potential bioavailability of the compounds screened and is summarized fully in Chapter 3.
ContributorsAlam, Mohammad Parvez (Author) / Hecht, Sidney M. (Thesis advisor) / Gould, Ian R (Committee member) / Moore, Ana (Committee member) / Arizona State University (Publisher)
Created2014
Description
Increasing concentrations of carbon dioxide in the atmosphere will inevitably lead to long-term changes in climate that can have serious consequences. Controlling anthropogenic emission of carbon dioxide into the atmosphere, however, represents a significant technological challenge. Various chemical approaches have been suggested, perhaps the most promising of these is based on electrochemical trapping of carbon dioxide using pyridine and derivatives. Optimization of this process requires a detailed understanding of the mechanisms of the reactions of reduced pyridines with carbon dioxide, which are not currently well known. This thesis describes a detailed mechanistic study of the nucleophilic and Bronsted basic properties of the radical anion of bipyridine as a model pyridine derivative, formed by one-electron reduction, with particular emphasis on the reactions with carbon dioxide. A time-resolved spectroscopic method was used to characterize the key intermediates and determine the kinetics of the reactions of the radical anion and its protonated radical form. Using a pulsed nanosecond laser, the bipyridine radical anion could be generated in-situ in less than 100 ns, which allows fast reactions to be monitored in real time. The bipyridine radical anion was found to be a very powerful one-electron donor, Bronsted base and nucleophile. It reacts by addition to the C=O bonds of ketones with a bimolecular rate constant around 1* 107 M-1 s-1. These are among the fastest nucleophilic additions that have been reported in literature. Temperature dependence studies demonstrate very low activation energies and large Arrhenius pre-exponential parameters, consistent with very high reactivity. The kinetics of E2 elimination, where the radical anion acts as a base, and SN2 substitution, where the radical anion acts as a nucleophile, are also characterized by large bimolecular rate constants in the range ca. 106 - 107 M-1 s-1. The pKa of the bipyridine radical anion was measured using a kinetic method and analysis of the data using a Marcus theory model for proton transfer. The bipyridine radical anion is found to have a pKa of 40±5 in DMSO. The reorganization energy for the proton transfer reaction was found to be 70±5 kJ/mol. The bipyridine radical anion was found to react very rapidly with carbon dioxide, with a bimolecular rate constant of 1* 108 M-1 s-1 and a small activation energy, whereas the protonated radical reacted with carbon dioxide with a rate constant that was too small to measure. The kinetic and thermodynamic data obtained in this work can be used to understand the mechanisms of the reactions of pyridines with carbon dioxide under reducing conditions.
ContributorsRanjan, Rajeev (Author) / Gould, Ian R (Thesis advisor) / Buttry, Daniel A (Thesis advisor) / Yarger, Jeff (Committee member) / Seo, Dong-Kyun (Committee member) / Arizona State University (Publisher)
Created2015
Description
Cellular redox phenomena are essential for the life of organisms. Described here is a summary of the synthesis of a number of redox-cycling therapeutic agents. The work centers on the synthesis of antitumor antibiotic bleomycin congeners. In addition, the synthesis of pyridinol analogues of alpha-tocopherol is also described. The bleomycins (BLMs) are a group of glycopeptide antibiotics that have been used clinically to treat several types of cancers. The antitumor activity of BLM is thought to be related to its degradation of DNA, and possibly RNA. Previous studies have indicated that the methylvalerate subunit of bleomycin plays an important role in facilitating DNA cleavage by bleomycin and deglycobleomycin. A series of methylvalerate analogues have been synthesized and incorporated into deglycobleomycin congeners by the use of solid-phase synthesis. All of the deglycobleomycin analogues were found to effect the relaxation of plasmid DNA. Those analogues having aromatic C4-substituents exhibited cleavage efficiency comparable to that of deglycoBLM A5. Some, but not all, of the deglycoBLM analogues were also capable of mediating sequence-selective DNA cleavage. The second project focused on the synthesis of bicyclic pyridinol analogues of alpha-tocopherol. Bicyclic pyridinol antioxidants have recently been reported to suppress the autoxidation of methyl linoleate more effectively than alpha-tocopherol. However, the complexity of the synthetic routes has hampered their further development as therapeutic agents. Described herein is a concise synthesis of two bicyclic pridinol antioxidants and a facile approach to their derivatives with simple alkyl chains attached to the antioxidant core. These analogues were shown to retain biological activity and exhibit tocopherol-like behaviour.
ContributorsCai, Xiaoqing (Author) / Hecht, Sidney M. (Thesis advisor) / Gould, Ian R (Committee member) / Hartnett, Hilairy E (Committee member) / Arizona State University (Publisher)
Created2011
Description
This dissertation examines two topics of emerging interest in the field of organic geochemistry. The topic of the first portion of the dissertation is cold organic geochemistry on Saturn's moon Titan. Titan has an atmosphere and surface that are rich in organic compounds. Liquid hydrocarbons exist on the surface, most famously as lakes. Photochemical reactions produce solid organics in Titan's atmosphere, and these materials settle onto the surface. At the surface, liquids can interact with solids, and geochemical processes can occur. To better understand these processes, I developed a thermodynamic model that can be used to calculate the solubilities of gases and solids in liquid hydrocarbons at cryogenic temperatures. The model was parameterized using experimental data, and provides a good fit to the data. Application of the model to Titan reveals that the equilibrium composition of surface liquids depends on the abundance of methane in the local atmosphere. The model also indicates that solid acetylene should be quite soluble in surface liquids, which implies that acetylene-rich rocks should be susceptible to chemical erosion, and acetylene evaporites may form on Titan. In the latter half of this dissertation, I focus on hot organic geochemistry below the surface of the Earth. Organic compounds are common in sediments. Burial of sediments leads to changes in physical and chemical conditions, promoting organic reactions. An important organic reaction in subsurface environments is decarboxylation, which generates hydrocarbons and carbon dioxide from simple organic acids. Fundamental knowledge about decarboxylation is required to better understand how the organic and inorganic compositions of sediments evolve in response to changing geochemical conditions. I performed experiments with the model compound phenylacetic acid to obtain information about mechanisms of decarboxylation in hydrothermal fluids. Patterns in rates of decarboxylation of substituted phenylacetic acids point to a mechanism that proceeds through a ring-protonated zwitterion of phenylacetic acid. In contrast, substituted sodium phenylacetates exhibit a different kinetic pattern, one that is consistent with the formation of the benzyl anion as an intermediate. Results from experiments with added hydrochloric acid or sodium hydroxide, and deuterated water agree with these interpretations. Thus, speciation dictates mechanism of decarboxylation.
ContributorsGlein, Christopher R (Author) / Shock, Everett L (Thesis advisor) / Hartnett, Hilairy E (Committee member) / Zolotov, Mikhail Y (Committee member) / Williams, Lynda B (Committee member) / Gould, Ian R (Committee member) / Arizona State University (Publisher)
Created2012
Description
In a collaborative environment where multiple robots and human beings are expected
to collaborate to perform a task, it becomes essential for a robot to be aware of multiple
agents working in its work environment. A robot must also learn to adapt to
different agents in the workspace and conduct its interaction based on the presence
of these agents. A theoretical framework was introduced which performs interaction
learning from demonstrations in a two-agent work environment, and it is called
Interaction Primitives.
This document is an in-depth description of the new state of the art Python
Framework for Interaction Primitives between two agents in a single as well as multiple
task work environment and extension of the original framework in a work environment
with multiple agents doing a single task. The original theory of Interaction
Primitives has been extended to create a framework which will capture correlation
between more than two agents while performing a single task. The new state of the
art Python framework is an intuitive, generic, easy to install and easy to use python
library which can be applied to use the Interaction Primitives framework in a work
environment. This library was tested in simulated environments and controlled laboratory
environment. The results and benchmarks of this library are available in the
related sections of this document.
to collaborate to perform a task, it becomes essential for a robot to be aware of multiple
agents working in its work environment. A robot must also learn to adapt to
different agents in the workspace and conduct its interaction based on the presence
of these agents. A theoretical framework was introduced which performs interaction
learning from demonstrations in a two-agent work environment, and it is called
Interaction Primitives.
This document is an in-depth description of the new state of the art Python
Framework for Interaction Primitives between two agents in a single as well as multiple
task work environment and extension of the original framework in a work environment
with multiple agents doing a single task. The original theory of Interaction
Primitives has been extended to create a framework which will capture correlation
between more than two agents while performing a single task. The new state of the
art Python framework is an intuitive, generic, easy to install and easy to use python
library which can be applied to use the Interaction Primitives framework in a work
environment. This library was tested in simulated environments and controlled laboratory
environment. The results and benchmarks of this library are available in the
related sections of this document.
ContributorsKumar, Ashish, M.S (Author) / Amor, Hani Ben (Thesis advisor) / Zhang, Yu (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2017
Description
Computer Vision as a eld has gone through signicant changes in the last decade.
The eld has seen tremendous success in designing learning systems with hand-crafted
features and in using representation learning to extract better features. In this dissertation
some novel approaches to representation learning and task learning are studied.
Multiple-instance learning which is generalization of supervised learning, is one
example of task learning that is discussed. In particular, a novel non-parametric k-
NN-based multiple-instance learning is proposed, which is shown to outperform other
existing approaches. This solution is applied to a diabetic retinopathy pathology
detection problem eectively.
In cases of representation learning, generality of neural features are investigated
rst. This investigation leads to some critical understanding and results in feature
generality among datasets. The possibility of learning from a mentor network instead
of from labels is then investigated. Distillation of dark knowledge is used to eciently
mentor a small network from a pre-trained large mentor network. These studies help
in understanding representation learning with smaller and compressed networks.
The eld has seen tremendous success in designing learning systems with hand-crafted
features and in using representation learning to extract better features. In this dissertation
some novel approaches to representation learning and task learning are studied.
Multiple-instance learning which is generalization of supervised learning, is one
example of task learning that is discussed. In particular, a novel non-parametric k-
NN-based multiple-instance learning is proposed, which is shown to outperform other
existing approaches. This solution is applied to a diabetic retinopathy pathology
detection problem eectively.
In cases of representation learning, generality of neural features are investigated
rst. This investigation leads to some critical understanding and results in feature
generality among datasets. The possibility of learning from a mentor network instead
of from labels is then investigated. Distillation of dark knowledge is used to eciently
mentor a small network from a pre-trained large mentor network. These studies help
in understanding representation learning with smaller and compressed networks.
ContributorsVenkatesan, Ragav (Author) / Li, Baoxin (Thesis advisor) / Turaga, Pavan (Committee member) / Yang, Yezhou (Committee member) / Davulcu, Hasan (Committee member) / Arizona State University (Publisher)
Created2017
Description
With the rise of the Big Data Era, an exponential amount of network data is being generated at an unprecedented rate across a wide-range of high impact micro and macro areas of research---from protein interaction to social networks. The critical challenge is translating this large scale network data into actionable information.
A key task in the data translation is the analysis of network connectivity via marked nodes---the primary focus of our research. We have developed a framework for analyzing network connectivity via marked nodes in large scale graphs, utilizing novel algorithms in three interrelated areas: (1) analysis of a single seed node via it’s ego-centric network (AttriPart algorithm); (2) pathway identification between two seed nodes (K-Simple Shortest Paths Multithreaded and Search Reduced (KSSPR) algorithm); and (3) tree detection, defining the interaction between three or more seed nodes (Shortest Path MST algorithm).
In an effort to address both fundamental and applied research issues, we have developed the LocalForcasting algorithm to explore how network connectivity analysis can be applied to local community evolution and recommender systems. The goal is to apply the LocalForecasting algorithm to various domains---e.g., friend suggestions in social networks or future collaboration in co-authorship networks. This algorithm utilizes link prediction in combination with the AttriPart algorithm to predict future connections in local graph partitions.
Results show that our proposed AttriPart algorithm finds up to 1.6x denser local partitions, while running approximately 43x faster than traditional local partitioning techniques (PageRank-Nibble). In addition, our LocalForecasting algorithm demonstrates a significant improvement in the number of nodes and edges correctly predicted over baseline methods. Furthermore, results for the KSSPR algorithm demonstrate a speed-up of up to 2.5x the standard k-simple shortest paths algorithm.
A key task in the data translation is the analysis of network connectivity via marked nodes---the primary focus of our research. We have developed a framework for analyzing network connectivity via marked nodes in large scale graphs, utilizing novel algorithms in three interrelated areas: (1) analysis of a single seed node via it’s ego-centric network (AttriPart algorithm); (2) pathway identification between two seed nodes (K-Simple Shortest Paths Multithreaded and Search Reduced (KSSPR) algorithm); and (3) tree detection, defining the interaction between three or more seed nodes (Shortest Path MST algorithm).
In an effort to address both fundamental and applied research issues, we have developed the LocalForcasting algorithm to explore how network connectivity analysis can be applied to local community evolution and recommender systems. The goal is to apply the LocalForecasting algorithm to various domains---e.g., friend suggestions in social networks or future collaboration in co-authorship networks. This algorithm utilizes link prediction in combination with the AttriPart algorithm to predict future connections in local graph partitions.
Results show that our proposed AttriPart algorithm finds up to 1.6x denser local partitions, while running approximately 43x faster than traditional local partitioning techniques (PageRank-Nibble). In addition, our LocalForecasting algorithm demonstrates a significant improvement in the number of nodes and edges correctly predicted over baseline methods. Furthermore, results for the KSSPR algorithm demonstrate a speed-up of up to 2.5x the standard k-simple shortest paths algorithm.
ContributorsFreitas, Scott (Author) / Tong, Hanghang (Thesis advisor) / Maciejewski, Ross (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2018
Description
The performance of most of the visual computing tasks depends on the quality of the features extracted from the raw data. Insightful feature representation increases the performance of many learning algorithms by exposing the underlying explanatory factors of the output for the unobserved input. A good representation should also handle anomalies in the data such as missing samples and noisy input caused by the undesired, external factors of variation. It should also reduce the data redundancy. Over the years, many feature extraction processes have been invented to produce good representations of raw images and videos.
The feature extraction processes can be categorized into three groups. The first group contains processes that are hand-crafted for a specific task. Hand-engineering features requires the knowledge of domain experts and manual labor. However, the feature extraction process is interpretable and explainable. Next group contains the latent-feature extraction processes. While the original feature lies in a high-dimensional space, the relevant factors for a task often lie on a lower dimensional manifold. The latent-feature extraction employs hidden variables to expose the underlying data properties that cannot be directly measured from the input. Latent features seek a specific structure such as sparsity or low-rank into the derived representation through sophisticated optimization techniques. The last category is that of deep features. These are obtained by passing raw input data with minimal pre-processing through a deep network. Its parameters are computed by iteratively minimizing a task-based loss.
In this dissertation, I present four pieces of work where I create and learn suitable data representations. The first task employs hand-crafted features to perform clinically-relevant retrieval of diabetic retinopathy images. The second task uses latent features to perform content-adaptive image enhancement. The third task ranks a pair of images based on their aestheticism. The goal of the last task is to capture localized image artifacts in small datasets with patch-level labels. For both these tasks, I propose novel deep architectures and show significant improvement over the previous state-of-art approaches. A suitable combination of feature representations augmented with an appropriate learning approach can increase performance for most visual computing tasks.
The feature extraction processes can be categorized into three groups. The first group contains processes that are hand-crafted for a specific task. Hand-engineering features requires the knowledge of domain experts and manual labor. However, the feature extraction process is interpretable and explainable. Next group contains the latent-feature extraction processes. While the original feature lies in a high-dimensional space, the relevant factors for a task often lie on a lower dimensional manifold. The latent-feature extraction employs hidden variables to expose the underlying data properties that cannot be directly measured from the input. Latent features seek a specific structure such as sparsity or low-rank into the derived representation through sophisticated optimization techniques. The last category is that of deep features. These are obtained by passing raw input data with minimal pre-processing through a deep network. Its parameters are computed by iteratively minimizing a task-based loss.
In this dissertation, I present four pieces of work where I create and learn suitable data representations. The first task employs hand-crafted features to perform clinically-relevant retrieval of diabetic retinopathy images. The second task uses latent features to perform content-adaptive image enhancement. The third task ranks a pair of images based on their aestheticism. The goal of the last task is to capture localized image artifacts in small datasets with patch-level labels. For both these tasks, I propose novel deep architectures and show significant improvement over the previous state-of-art approaches. A suitable combination of feature representations augmented with an appropriate learning approach can increase performance for most visual computing tasks.
ContributorsChandakkar, Parag Shridhar (Author) / Li, Baoxin (Thesis advisor) / Yang, Yezhou (Committee member) / Turaga, Pavan (Committee member) / Davulcu, Hasan (Committee member) / Arizona State University (Publisher)
Created2017