Matching Items (132)
Filtering by
- Genre: Academic theses
- Genre: Masters Thesis
- Creators: Li, Baoxin
- Member of: Theses and Dissertations
Description
In recent years, Convolutional Neural Networks (CNNs) have been widely used in not only the computer vision community but also within the medical imaging community. Specifically, the use of pre-trained CNNs on large-scale datasets (e.g., ImageNet) via transfer learning for a variety of medical imaging applications, has become the de facto standard within both communities.
However, to fit the current paradigm, 3D imaging tasks have to be reformulated and solved in 2D, losing rich 3D contextual information. Moreover, pre-trained models on natural images never see any biomedical images and do not have knowledge about anatomical structures present in medical images. To overcome the above limitations, this thesis proposes an image out-painting self-supervised proxy task to develop pre-trained models directly from medical images without utilizing systematic annotations. The idea is to randomly mask an image and train the model to predict the missing region. It is demonstrated that by predicting missing anatomical structures when seeing only parts of the image, the model will learn generic representation yielding better performance on various medical imaging applications via transfer learning.
The extensive experiments demonstrate that the proposed proxy task outperforms training from scratch in six out of seven medical imaging applications covering 2D and 3D classification and segmentation. Moreover, image out-painting proxy task offers competitive performance to state-of-the-art models pre-trained on ImageNet and other self-supervised baselines such as in-painting. Owing to its outstanding performance, out-painting is utilized as one of the self-supervised proxy tasks to provide generic 3D pre-trained models for medical image analysis.
However, to fit the current paradigm, 3D imaging tasks have to be reformulated and solved in 2D, losing rich 3D contextual information. Moreover, pre-trained models on natural images never see any biomedical images and do not have knowledge about anatomical structures present in medical images. To overcome the above limitations, this thesis proposes an image out-painting self-supervised proxy task to develop pre-trained models directly from medical images without utilizing systematic annotations. The idea is to randomly mask an image and train the model to predict the missing region. It is demonstrated that by predicting missing anatomical structures when seeing only parts of the image, the model will learn generic representation yielding better performance on various medical imaging applications via transfer learning.
The extensive experiments demonstrate that the proposed proxy task outperforms training from scratch in six out of seven medical imaging applications covering 2D and 3D classification and segmentation. Moreover, image out-painting proxy task offers competitive performance to state-of-the-art models pre-trained on ImageNet and other self-supervised baselines such as in-painting. Owing to its outstanding performance, out-painting is utilized as one of the self-supervised proxy tasks to provide generic 3D pre-trained models for medical image analysis.
ContributorsSodha, Vatsal Arvindkumar (Author) / Liang, Jianming (Thesis advisor) / Devarakonda, Murthy (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2020
Description
Convolutional Neural Network (CNN) has achieved state-of-the-art performance in numerous applications like computer vision, natural language processing, robotics etc. The advancement of High-Performance Computing systems equipped with dedicated hardware accelerators has also paved the way towards the success of compute intensive CNNs. Graphics Processing Units (GPUs), with massive processing capability, have been of general interest for the acceleration of CNNs. Recently, Field Programmable Gate Arrays (FPGAs) have been promising in CNN acceleration since they offer high performance while also being re-configurable to support the evolution of CNNs. This work focuses on a design methodology to accelerate CNNs on FPGA with low inference latency and high-throughput which are crucial for scenarios like self-driving cars, video surveillance etc. It also includes optimizations which reduce the resource utilization by a large margin with a small degradation in performance thus making the design suitable for low-end FPGA devices as well.
FPGA accelerators often suffer due to the limited main memory bandwidth. Also, highly parallel designs with large resource utilization often end up achieving low operating frequency due to poor routing. This work employs data fetch and buffer mechanisms, designed specifically for the memory access pattern of CNNs, that overlap computation with memory access. This work proposes a novel arrangement of the systolic processing element array to achieve high frequency and consume less resources than the existing works. Also, support has been extended to more complicated CNNs to do video processing. On Intel Arria 10 GX1150, the design operates at a frequency as high as 258MHz and performs single inference of VGG-16 and C3D in 23.5ms and 45.6ms respectively. For VGG-16 and C3D the design offers a throughput of 66.1 and 23.98 inferences/s respectively. This design can outperform other FPGA 2D CNN accelerators by up to 9.7 times and 3D CNN accelerators by up to 2.7 times.
FPGA accelerators often suffer due to the limited main memory bandwidth. Also, highly parallel designs with large resource utilization often end up achieving low operating frequency due to poor routing. This work employs data fetch and buffer mechanisms, designed specifically for the memory access pattern of CNNs, that overlap computation with memory access. This work proposes a novel arrangement of the systolic processing element array to achieve high frequency and consume less resources than the existing works. Also, support has been extended to more complicated CNNs to do video processing. On Intel Arria 10 GX1150, the design operates at a frequency as high as 258MHz and performs single inference of VGG-16 and C3D in 23.5ms and 45.6ms respectively. For VGG-16 and C3D the design offers a throughput of 66.1 and 23.98 inferences/s respectively. This design can outperform other FPGA 2D CNN accelerators by up to 9.7 times and 3D CNN accelerators by up to 2.7 times.
ContributorsRavi, Pravin Kumar (Author) / Zhao, Ming (Thesis advisor) / Li, Baoxin (Committee member) / Ren, Fengbo (Committee member) / Arizona State University (Publisher)
Created2020
Description
Access to real-time situational information including the relative position and motion of surrounding objects is critical for safe and independent travel. Object or obstacle (OO) detection at a distance is primarily a task of the visual system due to the high resolution information the eyes are able to receive from afar. As a sensory organ in particular, the eyes have an unparalleled ability to adjust to varying degrees of light, color, and distance. Therefore, in the case of a non-visual traveler, someone who is blind or low vision, access to visual information is unattainable if it is positioned beyond the reach of the preferred mobility device or outside the path of travel. Although, the area of assistive technology in terms of electronic travel aids (ETA’s) has received considerable attention over the last two decades; surprisingly, the field has seen little work in the area focused on augmenting rather than replacing current non-visual travel techniques, methods, and tools. Consequently, this work describes the design of an intuitive tactile language and series of wearable tactile interfaces (the Haptic Chair, HaptWrap, and HapBack) to deliver real-time spatiotemporal data. The overall intuitiveness of the haptic mappings conveyed through the tactile interfaces are evaluated using a combination of absolute identification accuracy of a series of patterns and subjective feedback through post-experiment surveys. Two types of spatiotemporal representations are considered: static patterns representing object location at a single time instance, and dynamic patterns, added in the HaptWrap, which represent object movement over a time interval. Results support the viability of multi-dimensional haptics applied to the body to yield an intuitive understanding of dynamic interactions occurring around the navigator during travel. Lastly, it is important to point out that the guiding principle of this work centered on providing the navigator with spatial knowledge otherwise unattainable through current mobility techniques, methods, and tools, thus, providing the \emph{navigator} with the information necessary to make informed navigation decisions independently, at a distance.
ContributorsDuarte, Bryan Joiner (Author) / McDaniel, Troy (Thesis advisor) / Davulcu, Hasan (Committee member) / Li, Baoxin (Committee member) / Venkateswara, Hemanth (Committee member) / Arizona State University (Publisher)
Created2020
Description
Image super-resolution (SR) is a low-level image processing task, which has manyapplications such as medical imaging, satellite image processing, and video enhancement,
etc. Given a low resolution image, it aims to reconstruct a high resolution
image. The problem is ill-posed since there can be more than one high resolution
image corresponding to the same low-resolution image. To address this problem, a
number of machine learning-based approaches have been proposed.
In this dissertation, I present my works on single image super-resolution (SISR)
and accelerated magnetic resonance imaging (MRI) (a.k.a. super-resolution on MR
images), followed by the investigation on transfer learning for accelerated MRI reconstruction.
For the SISR, a dictionary-based approach and two reconstruction based
approaches are presented. To be precise, a convex dictionary learning (CDL)
algorithm is proposed by constraining the dictionary atoms to be formed by nonnegative
linear combination of the training data, which is a natural, desired property.
Also, two reconstruction-based single methods are presented, which make use
of (i)the joint regularization, where a group-residual-based regularization (GRR) and
a ridge-regression-based regularization (3R) are combined; (ii)the collaborative representation
and non-local self-similarity. After that, two deep learning approaches
are proposed, aiming at reconstructing high-quality images from accelerated MRI
acquisition. Residual Dense Block (RDB) and feedback connection are introduced
in the proposed models. In the last chapter, the feasibility of transfer learning for
accelerated MRI reconstruction is discussed.
etc. Given a low resolution image, it aims to reconstruct a high resolution
image. The problem is ill-posed since there can be more than one high resolution
image corresponding to the same low-resolution image. To address this problem, a
number of machine learning-based approaches have been proposed.
In this dissertation, I present my works on single image super-resolution (SISR)
and accelerated magnetic resonance imaging (MRI) (a.k.a. super-resolution on MR
images), followed by the investigation on transfer learning for accelerated MRI reconstruction.
For the SISR, a dictionary-based approach and two reconstruction based
approaches are presented. To be precise, a convex dictionary learning (CDL)
algorithm is proposed by constraining the dictionary atoms to be formed by nonnegative
linear combination of the training data, which is a natural, desired property.
Also, two reconstruction-based single methods are presented, which make use
of (i)the joint regularization, where a group-residual-based regularization (GRR) and
a ridge-regression-based regularization (3R) are combined; (ii)the collaborative representation
and non-local self-similarity. After that, two deep learning approaches
are proposed, aiming at reconstructing high-quality images from accelerated MRI
acquisition. Residual Dense Block (RDB) and feedback connection are introduced
in the proposed models. In the last chapter, the feasibility of transfer learning for
accelerated MRI reconstruction is discussed.
ContributorsDing, Pak Lun Kevin (Author) / Li, Baoxin (Thesis advisor) / Wu, Teresa (Committee member) / Wang, Yalin (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2020
Description
Over the past decade, machine learning research has made great strides and significant impact in several fields. Its success is greatly attributed to the development of effective machine learning algorithms like deep neural networks (a.k.a. deep learning), availability of large-scale databases and access to specialized hardware like Graphic Processing Units. When designing and training machine learning systems, researchers often assume access to large quantities of data that capture different possible variations. Variations in the data is needed to incorporate desired invariance and robustness properties in the machine learning system, especially in the case of deep learning algorithms. However, it is very difficult to gather such data in a real-world setting. For example, in certain medical/healthcare applications, it is very challenging to have access to data from all possible scenarios or with the necessary amount of variations as required to train the system. Additionally, the over-parameterized and unconstrained nature of deep neural networks can cause them to be poorly trained and in many cases over-confident which, in turn, can hamper their reliability and generalizability. This dissertation is a compendium of my research efforts to address the above challenges. I propose building invariant feature representations by wedding concepts from topological data analysis and Riemannian geometry, that automatically incorporate the desired invariance properties for different computer vision applications. I discuss how deep learning can be used to address some of the common challenges faced when working with topological data analysis methods. I describe alternative learning strategies based on unsupervised learning and transfer learning to address issues like dataset shifts and limited training data. Finally, I discuss my preliminary work on applying simple orthogonal constraints on deep learning feature representations to help develop more reliable and better calibrated models.
ContributorsSom, Anirudh (Author) / Turaga, Pavan (Thesis advisor) / Krishnamurthi, Narayanan (Committee member) / Spanias, Andreas (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2020
Description
Technological advances have allowed for the assimilation of a variety of data, driving a shift away from the use of simpler and constrained patterns to more complex and diverse patterns in retrieval and analysis of such data. This shift has inundated the conventional techniques and has stressed the need for intelligent mechanisms that can model the complex patterns in the data. Deep neural networks have shown some success at capturing complex patterns, including the so-called attentioned networks, have significant shortcomings in distinguishing what is important in data from what is noise. This dissertation observes that the traditional neural networks primarily rely solely on gradient-based learning to model deep features maps while ignoring the key insight in the data that can be leveraged as complementary information to help learn an accurate model. In particular, this dissertation shows that the localized multi-scale features (captured implicitly or explicitly) can be leveraged to help improve model performance as these features capture salient informative points in the data.
This dissertation focuses on “working with the data, not just on data”, i.e. leveraging feature saliency through pre-training, in-training, and post-training analysis of the data. In particular, non-neural localized multi-scale feature extraction, in images and time series, are relatively cheap to obtain and can provide a rough overview of the patterns in the data. Furthermore, localized features coupled with deep features can help learn a high performing network. A pre-training analysis of sizes, complexities, and distribution of these localized features can help intelligently allocate a user-provided kernel budget in the network as a single-shot hyper-parameter search. Additionally, these localized features can be used as a secondary input modality to the network for cross-attention. Retraining pre-trained networks can be a costly process, yet, a post-training analysis of model inferences can allow for learning the importance of individual network parameters to the model inferences thus facilitating a retraining-free network sparsification with minimal impact on the model performance. Furthermore, effective in-training analysis of the intermediate features in the network help learn the importance of individual intermediate features (neural attention) and this analysis can be achieved through simulating local-extrema detection or learning features simultaneously and understanding their co-occurrences. In summary, this dissertation argues and establishes that, if appropriately leveraged, localized features and their feature saliency can help learn high-accurate, yet cheaper networks.
This dissertation focuses on “working with the data, not just on data”, i.e. leveraging feature saliency through pre-training, in-training, and post-training analysis of the data. In particular, non-neural localized multi-scale feature extraction, in images and time series, are relatively cheap to obtain and can provide a rough overview of the patterns in the data. Furthermore, localized features coupled with deep features can help learn a high performing network. A pre-training analysis of sizes, complexities, and distribution of these localized features can help intelligently allocate a user-provided kernel budget in the network as a single-shot hyper-parameter search. Additionally, these localized features can be used as a secondary input modality to the network for cross-attention. Retraining pre-trained networks can be a costly process, yet, a post-training analysis of model inferences can allow for learning the importance of individual network parameters to the model inferences thus facilitating a retraining-free network sparsification with minimal impact on the model performance. Furthermore, effective in-training analysis of the intermediate features in the network help learn the importance of individual intermediate features (neural attention) and this analysis can be achieved through simulating local-extrema detection or learning features simultaneously and understanding their co-occurrences. In summary, this dissertation argues and establishes that, if appropriately leveraged, localized features and their feature saliency can help learn high-accurate, yet cheaper networks.
ContributorsGarg, Yash (Author) / Candan, K. Selcuk (Thesis advisor) / Davulcu, Hasan (Committee member) / Li, Baoxin (Committee member) / Sapino, Maria Luisa (Committee member) / Arizona State University (Publisher)
Created2020
Description
The burden of adaptation has been a major limiting factor in the adoption rates of new wearable assistive technologies. This burden has created a necessity for the exploration and combination of two key concepts in the development of upcoming wearables: anticipation and invisibility. The combination of these two topics has created the field of Anticipatory and Invisible Interfaces (AII)
In this dissertation, a novel framework is introduced for the development of anticipatory devices that augment the proprioceptive system in individuals with neurodegenerative disorders in a seamless way that scaffolds off of existing cognitive feedback models. The framework suggests three main categories of consideration in the development of devices which are anticipatory and invisible:
• Idiosyncratic Design: How do can a design encapsulate the unique characteristics of the individual in the design of assistive aids?
• Adaptation to Intrapersonal Variations: As individuals progress through the various stages of a disability
eurological disorder, how can the technology adapt thresholds for feedback over time to address these shifts in ability?
• Context Aware Invisibility: How can the mechanisms of interaction be modified in order to reduce cognitive load?
The concepts proposed in this framework can be generalized to a broad range of domains; however, there are two primary applications for this work: rehabilitation and assistive aids. In preliminary studies, the framework is applied in the areas of Parkinsonian freezing of gait anticipation and the anticipation of body non-compliance during rehabilitative exercise.
In this dissertation, a novel framework is introduced for the development of anticipatory devices that augment the proprioceptive system in individuals with neurodegenerative disorders in a seamless way that scaffolds off of existing cognitive feedback models. The framework suggests three main categories of consideration in the development of devices which are anticipatory and invisible:
• Idiosyncratic Design: How do can a design encapsulate the unique characteristics of the individual in the design of assistive aids?
• Adaptation to Intrapersonal Variations: As individuals progress through the various stages of a disability
eurological disorder, how can the technology adapt thresholds for feedback over time to address these shifts in ability?
• Context Aware Invisibility: How can the mechanisms of interaction be modified in order to reduce cognitive load?
The concepts proposed in this framework can be generalized to a broad range of domains; however, there are two primary applications for this work: rehabilitation and assistive aids. In preliminary studies, the framework is applied in the areas of Parkinsonian freezing of gait anticipation and the anticipation of body non-compliance during rehabilitative exercise.
ContributorsTadayon, Arash (Author) / Panchanathan, Sethuraman (Thesis advisor) / McDaniel, Troy (Committee member) / Krishnamurthi, Narayanan (Committee member) / Davulcu, Hasan (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2020
Description
The development of the internet provided new means for people to communicate effectively and share their ideas. There has been a decline in the consumption of newspapers and traditional broadcasting media toward online social mediums in recent years. Social media has been introduced as a new way of increasing democratic discussions on political and social matters. Among social media, Twitter is widely used by politicians, government officials, communities, and parties to make announcements and reach their voice to their followers. This greatly increases the acceptance domain of the medium.
The usage of social media during social and political campaigns has been the subject of a lot of social science studies including the Occupy Wall Street movement, The Arab Spring, the United States (US) election, more recently The Brexit campaign. The wide
spread usage of social media in this space and the active participation of people in the discussions on social media made this communication channel a suitable place for spreading propaganda to alter public opinion.
An interesting feature of twitter is the feasibility of which bots can be programmed to operate on this platform. Social media bots are automated agents engineered to emulate the activity of a human being by tweeting some specific content, replying to users, magnifying certain topics by retweeting them. Network on these bots is called botnets and describing the collaboration of connected computers with programs that communicates across multiple devices to perform some task.
In this thesis, I will study how bots can influence the opinion, finding which parameters are playing a role in shrinking or coalescing the communities, and finally logically proving the effectiveness of each of the hypotheses.
The usage of social media during social and political campaigns has been the subject of a lot of social science studies including the Occupy Wall Street movement, The Arab Spring, the United States (US) election, more recently The Brexit campaign. The wide
spread usage of social media in this space and the active participation of people in the discussions on social media made this communication channel a suitable place for spreading propaganda to alter public opinion.
An interesting feature of twitter is the feasibility of which bots can be programmed to operate on this platform. Social media bots are automated agents engineered to emulate the activity of a human being by tweeting some specific content, replying to users, magnifying certain topics by retweeting them. Network on these bots is called botnets and describing the collaboration of connected computers with programs that communicates across multiple devices to perform some task.
In this thesis, I will study how bots can influence the opinion, finding which parameters are playing a role in shrinking or coalescing the communities, and finally logically proving the effectiveness of each of the hypotheses.
ContributorsAhmadi, Mohsen (Author) / Davulcu, Hasan (Thesis advisor) / Sen, Arunabha (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2020
Description
Due to the advent of easy-to-use, portable, and cost-effective brain signal sensing devices, pervasive Brain-Machine Interface (BMI) applications using Electroencephalogram (EEG) are growing rapidly. The main objectives of these applications are: 1) pervasive collection of brain data from multiple users, 2) processing the collected data to recognize the corresponding mental states, and 3) providing real-time feedback to the end users, activating an actuator, or information harvesting by enterprises for further services. Developing BMI applications faces several challenges, such as cumbersome setup procedure, low signal-to-noise ratio, insufficient signal samples for analysis, and long processing times. Internet-of-Things (IoT) technologies provide the opportunity to solve these challenges through large scale data collection, fast data transmission, and computational offloading.
This research proposes an IoT-based framework, called BraiNet, that provides a standard design methodology for fulfilling the pervasive BMI applications requirements including: accuracy, timeliness, energy-efficiency, security, and dependability. BraiNet applies Machine Learning (ML) based solutions (e.g. classifiers and predictive models) to: 1) improve the accuracy of mental state detection on-the-go, 2) provide real-time feedback to the users, and 3) save power on mobile platforms. However, BraiNet inherits security breaches of IoT, due to applying off-the-shelf soft/hardware, high accessibility, and massive network size. ML algorithms, as the core technology for mental state recognition, are among the main targets for cyber attackers. Novel ML security solutions are proposed and added to BraiNet, which provide analytical methodologies for tuning the ML hyper-parameters to be secure against attacks.
To implement these solutions, two main optimization problems are solved: 1) maximizing accuracy, while minimizing delays and power consumption, and 2) maximizing the ML security, while keeping its accuracy high. Deep learning algorithms, delay and power models are developed to solve the former problem, while gradient-free optimization techniques, such as Bayesian optimization are applied for the latter. To test the framework, several BMI applications are implemented, such as EEG-based drivers fatigue detector (SafeDrive), EEG-based identification and authentication system (E-BIAS), and interactive movies that adapt to viewers mental states (nMovie). The results from the experiments on the implemented applications show the successful design of pervasive BMI applications based on the BraiNet framework.
This research proposes an IoT-based framework, called BraiNet, that provides a standard design methodology for fulfilling the pervasive BMI applications requirements including: accuracy, timeliness, energy-efficiency, security, and dependability. BraiNet applies Machine Learning (ML) based solutions (e.g. classifiers and predictive models) to: 1) improve the accuracy of mental state detection on-the-go, 2) provide real-time feedback to the users, and 3) save power on mobile platforms. However, BraiNet inherits security breaches of IoT, due to applying off-the-shelf soft/hardware, high accessibility, and massive network size. ML algorithms, as the core technology for mental state recognition, are among the main targets for cyber attackers. Novel ML security solutions are proposed and added to BraiNet, which provide analytical methodologies for tuning the ML hyper-parameters to be secure against attacks.
To implement these solutions, two main optimization problems are solved: 1) maximizing accuracy, while minimizing delays and power consumption, and 2) maximizing the ML security, while keeping its accuracy high. Deep learning algorithms, delay and power models are developed to solve the former problem, while gradient-free optimization techniques, such as Bayesian optimization are applied for the latter. To test the framework, several BMI applications are implemented, such as EEG-based drivers fatigue detector (SafeDrive), EEG-based identification and authentication system (E-BIAS), and interactive movies that adapt to viewers mental states (nMovie). The results from the experiments on the implemented applications show the successful design of pervasive BMI applications based on the BraiNet framework.
ContributorsSadeghi Oskooyee, Seyed Koosha (Author) / Gupta, Sandeep K S (Thesis advisor) / Santello, Marco (Committee member) / Li, Baoxin (Committee member) / Venkatasubramanian, Krishna K (Committee member) / Banerjee, Ayan (Committee member) / Arizona State University (Publisher)
Created2020
Description
This thesis introduces new techniques for clustering distributional data according to their geometric similarities. This work builds upon the optimal transportation (OT) problem that seeks global minimum cost for matching distributional data and leverages the connection between OT and power diagrams to solve different clustering problems. The OT formulation is based on the variational principle to differentiate hard cluster assignments, which was missing in the literature. This thesis shows multiple techniques to regularize and generalize OT to cope with various tasks including clustering, aligning, and interpolating distributional data. It also discusses the connections of the new formulation to other OT and clustering formulations to better understand their gaps and the means to close them. Finally, this thesis demonstrates the advantages of the proposed OT techniques in solving machine learning problems and their downstream applications in computer graphics, computer vision, and image processing.
ContributorsMi, Liang (Author) / Wang, Yalin (Thesis advisor) / Chen, Kewei (Committee member) / Karam, Lina (Committee member) / Li, Baoxin (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2020