Theses and Dissertations
This collection includes both ASU Theses and Dissertations, submitted by graduate students, and the Barrett, Honors College theses submitted by undergraduate students.
Displaying 1 - 10 of 16
Filtering by
- All Subjects: deep learning
- All Subjects: Music
- Creators: Turaga, Pavan
Description
Digital sound synthesis allows the creation of a great variety of sounds. Focusing on interesting or ecologically valid sounds for music, simulation, aesthetics, or other purposes limits the otherwise vast digital audio palette. Tools for creating such sounds vary from arbitrary methods of altering recordings to precise simulations of vibrating objects. In this work, methods of sound synthesis by re-sonification are considered. Re-sonification, herein, refers to the general process of analyzing, possibly transforming, and resynthesizing or reusing recorded sounds in meaningful ways, to convey information. Applied to soundscapes, re-sonification is presented as a means of conveying activity within an environment. Applied to the sounds of objects, this work examines modeling the perception of objects as well as their physical properties and the ability to simulate interactive events with such objects. To create soundscapes to re-sonify geographic environments, a method of automated soundscape design is presented. Using recorded sounds that are classified based on acoustic, social, semantic, and geographic information, this method produces stochastically generated soundscapes to re-sonify selected geographic areas. Drawing on prior knowledge, local sounds and those deemed similar comprise a locale's soundscape. In the context of re-sonifying events, this work examines processes for modeling and estimating the excitations of sounding objects. These include plucking, striking, rubbing, and any interaction that imparts energy into a system, affecting the resultant sound. A method of estimating a linear system's input, constrained to a signal-subspace, is presented and applied toward improving the estimation of percussive excitations for re-sonification. To work toward robust recording-based modeling and re-sonification of objects, new implementations of banded waveguide (BWG) models are proposed for object modeling and sound synthesis. Previous implementations of BWGs use arbitrary model parameters and may produce a range of simulations that do not match digital waveguide or modal models of the same design. Subject to linear excitations, some models proposed here behave identically to other equivalently designed physical models. Under nonlinear interactions, such as bowing, many of the proposed implementations exhibit improvements in the attack characteristics of synthesized sounds.
ContributorsFink, Alex M (Author) / Spanias, Andreas S (Thesis advisor) / Cook, Perry R. (Committee member) / Turaga, Pavan (Committee member) / Tsakalis, Konstantinos (Committee member) / Arizona State University (Publisher)
Created2013
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
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
Description
Mixture of experts is a machine learning ensemble approach that consists of individual models that are trained to be ``experts'' on subsets of the data, and a gating network that provides weights to output a combination of the expert predictions. Mixture of experts models do not currently see wide use due to difficulty in training diverse experts and high computational requirements. This work presents modifications of the mixture of experts formulation that use domain knowledge to improve training, and incorporate parameter sharing among experts to reduce computational requirements.
First, this work presents an application of mixture of experts models for quality robust visual recognition. First it is shown that human subjects outperform deep neural networks on classification of distorted images, and then propose a model, MixQualNet, that is more robust to distortions. The proposed model consists of ``experts'' that are trained on a particular type of image distortion. The final output of the model is a weighted sum of the expert models, where the weights are determined by a separate gating network. The proposed model also incorporates weight sharing to reduce the number of parameters, as well as increase performance.
Second, an application of mixture of experts to predict visual saliency is presented. A computational saliency model attempts to predict where humans will look in an image. In the proposed model, each expert network is trained to predict saliency for a set of closely related images. The final saliency map is computed as a weighted mixture of the expert networks' outputs, with weights determined by a separate gating network. The proposed model achieves better performance than several other visual saliency models and a baseline non-mixture model.
Finally, this work introduces a saliency model that is a weighted mixture of models trained for different levels of saliency. Levels of saliency include high saliency, which corresponds to regions where almost all subjects look, and low saliency, which corresponds to regions where some, but not all subjects look. The weighted mixture shows improved performance compared with baseline models because of the diversity of the individual model predictions.
First, this work presents an application of mixture of experts models for quality robust visual recognition. First it is shown that human subjects outperform deep neural networks on classification of distorted images, and then propose a model, MixQualNet, that is more robust to distortions. The proposed model consists of ``experts'' that are trained on a particular type of image distortion. The final output of the model is a weighted sum of the expert models, where the weights are determined by a separate gating network. The proposed model also incorporates weight sharing to reduce the number of parameters, as well as increase performance.
Second, an application of mixture of experts to predict visual saliency is presented. A computational saliency model attempts to predict where humans will look in an image. In the proposed model, each expert network is trained to predict saliency for a set of closely related images. The final saliency map is computed as a weighted mixture of the expert networks' outputs, with weights determined by a separate gating network. The proposed model achieves better performance than several other visual saliency models and a baseline non-mixture model.
Finally, this work introduces a saliency model that is a weighted mixture of models trained for different levels of saliency. Levels of saliency include high saliency, which corresponds to regions where almost all subjects look, and low saliency, which corresponds to regions where some, but not all subjects look. The weighted mixture shows improved performance compared with baseline models because of the diversity of the individual model predictions.
ContributorsDodge, Samuel Fuller (Author) / Karam, Lina (Thesis advisor) / Jayasuriya, Suren (Committee member) / Li, Baoxin (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2018
Description
Non-line-of-sight (NLOS) imaging of objects not visible to either the camera or illumina-
tion source is a challenging task with vital applications including surveillance and robotics.
Recent NLOS reconstruction advances have been achieved using time-resolved measure-
ments. Acquiring these time-resolved measurements requires expensive and specialized
detectors and laser sources. In work proposes a data-driven approach for NLOS 3D local-
ization requiring only a conventional camera and projector. The localisation is performed
using a voxelisation and a regression problem. Accuracy of greater than 90% is achieved
in localizing a NLOS object to a 5cm × 5cm × 5cm volume in real data. By adopting
the regression approach an object of width 10cm to localised to approximately 1.5cm. To
generalize to line-of-sight (LOS) scenes with non-planar surfaces, an adaptive lighting al-
gorithm is adopted. This algorithm, based on radiosity, identifies and illuminates scene
patches in the LOS which most contribute to the NLOS light paths, and can factor in sys-
tem power constraints. Improvements ranging from 6%-15% in accuracy with a non-planar
LOS wall using adaptive lighting is reported, demonstrating the advantage of combining
the physics of light transport with active illumination for data-driven NLOS imaging.
tion source is a challenging task with vital applications including surveillance and robotics.
Recent NLOS reconstruction advances have been achieved using time-resolved measure-
ments. Acquiring these time-resolved measurements requires expensive and specialized
detectors and laser sources. In work proposes a data-driven approach for NLOS 3D local-
ization requiring only a conventional camera and projector. The localisation is performed
using a voxelisation and a regression problem. Accuracy of greater than 90% is achieved
in localizing a NLOS object to a 5cm × 5cm × 5cm volume in real data. By adopting
the regression approach an object of width 10cm to localised to approximately 1.5cm. To
generalize to line-of-sight (LOS) scenes with non-planar surfaces, an adaptive lighting al-
gorithm is adopted. This algorithm, based on radiosity, identifies and illuminates scene
patches in the LOS which most contribute to the NLOS light paths, and can factor in sys-
tem power constraints. Improvements ranging from 6%-15% in accuracy with a non-planar
LOS wall using adaptive lighting is reported, demonstrating the advantage of combining
the physics of light transport with active illumination for data-driven NLOS imaging.
ContributorsChandran, Sreenithy (Author) / Jayasuriya, Suren (Thesis advisor) / Turaga, Pavan (Committee member) / Dasarathy, Gautam (Committee member) / Arizona State University (Publisher)
Created2019
Description
The tradition of building musical robots and automata is thousands of years old. Despite this rich history, even today musical robots do not play with as much nuance and subtlety as human musicians. In particular, most instruments allow the player to manipulate timbre while playing; if a violinist is told to sustain an E, they will select which string to play it on, how much bow pressure and velocity to use, whether to use the entire bow or only the portion near the tip or the frog, how close to the bridge or fingerboard to contact the string, whether or not to use a mute, and so forth. Each one of these choices affects the resulting timbre, and navigating this timbre space is part of the art of playing the instrument. Nonetheless, this type of timbral nuance has been largely ignored in the design of musical robots. Therefore, this dissertation introduces a suite of techniques that deal with timbral nuance in musical robots. Chapter 1 provides the motivating ideas and introduces Kiki, a robot designed by the author to explore timbral nuance. Chapter 2 provides a long history of musical robots, establishing the under-researched nature of timbral nuance. Chapter 3 is a comprehensive treatment of dynamic timbre production in percussion robots and, using Kiki as a case-study, provides a variety of techniques for designing striking mechanisms that produce a range of timbres similar to those produced by human players. Chapter 4 introduces a machine-learning algorithm for recognizing timbres, so that a robot can transcribe timbres played by a human during live performance. Chapter 5 introduces a technique that allows a robot to learn how to produce isolated instances of particular timbres by listening to a human play an examples of those timbres. The 6th and final chapter introduces a method that allows a robot to learn the musical context of different timbres; this is done in realtime during interactive improvisation between a human and robot, wherein the robot builds a statistical model of which timbres the human plays in which contexts, and uses this to inform its own playing.
ContributorsKrzyzaniak, Michael Joseph (Author) / Coleman, Grisha (Thesis advisor) / Turaga, Pavan (Committee member) / Artemiadis, Panagiotis (Committee member) / Arizona State University (Publisher)
Created2016
Description
High-level inference tasks in video applications such as recognition, video retrieval, and zero-shot classification have become an active research area in recent years. One fundamental requirement for such applications is to extract high-quality features that maintain high-level information in the videos.
Many video feature extraction algorithms have been purposed, such as STIP, HOG3D, and Dense Trajectories. These algorithms are often referred to as “handcrafted” features as they were deliberately designed based on some reasonable considerations. However, these algorithms may fail when dealing with high-level tasks or complex scene videos. Due to the success of using deep convolution neural networks (CNNs) to extract global representations for static images, researchers have been using similar techniques to tackle video contents. Typical techniques first extract spatial features by processing raw images using deep convolution architectures designed for static image classifications. Then simple average, concatenation or classifier-based fusion/pooling methods are applied to the extracted features. I argue that features extracted in such ways do not acquire enough representative information since videos, unlike images, should be characterized as a temporal sequence of semantically coherent visual contents and thus need to be represented in a manner considering both semantic and spatio-temporal information.
In this thesis, I propose a novel architecture to learn semantic spatio-temporal embedding for videos to support high-level video analysis. The proposed method encodes video spatial and temporal information separately by employing a deep architecture consisting of two channels of convolutional neural networks (capturing appearance and local motion) followed by their corresponding Fully Connected Gated Recurrent Unit (FC-GRU) encoders for capturing longer-term temporal structure of the CNN features. The resultant spatio-temporal representation (a vector) is used to learn a mapping via a Fully Connected Multilayer Perceptron (FC-MLP) to the word2vec semantic embedding space, leading to a semantic interpretation of the video vector that supports high-level analysis. I evaluate the usefulness and effectiveness of this new video representation by conducting experiments on action recognition, zero-shot video classification, and semantic video retrieval (word-to-video) retrieval, using the UCF101 action recognition dataset.
Many video feature extraction algorithms have been purposed, such as STIP, HOG3D, and Dense Trajectories. These algorithms are often referred to as “handcrafted” features as they were deliberately designed based on some reasonable considerations. However, these algorithms may fail when dealing with high-level tasks or complex scene videos. Due to the success of using deep convolution neural networks (CNNs) to extract global representations for static images, researchers have been using similar techniques to tackle video contents. Typical techniques first extract spatial features by processing raw images using deep convolution architectures designed for static image classifications. Then simple average, concatenation or classifier-based fusion/pooling methods are applied to the extracted features. I argue that features extracted in such ways do not acquire enough representative information since videos, unlike images, should be characterized as a temporal sequence of semantically coherent visual contents and thus need to be represented in a manner considering both semantic and spatio-temporal information.
In this thesis, I propose a novel architecture to learn semantic spatio-temporal embedding for videos to support high-level video analysis. The proposed method encodes video spatial and temporal information separately by employing a deep architecture consisting of two channels of convolutional neural networks (capturing appearance and local motion) followed by their corresponding Fully Connected Gated Recurrent Unit (FC-GRU) encoders for capturing longer-term temporal structure of the CNN features. The resultant spatio-temporal representation (a vector) is used to learn a mapping via a Fully Connected Multilayer Perceptron (FC-MLP) to the word2vec semantic embedding space, leading to a semantic interpretation of the video vector that supports high-level analysis. I evaluate the usefulness and effectiveness of this new video representation by conducting experiments on action recognition, zero-shot video classification, and semantic video retrieval (word-to-video) retrieval, using the UCF101 action recognition dataset.
ContributorsHu, Sheng-Hung (Author) / Li, Baoxin (Thesis advisor) / Turaga, Pavan (Committee member) / Liang, Jianming (Committee member) / Tong, Hanghang (Committee member) / Arizona State University (Publisher)
Created2016
Description
Compressive sensing theory allows to sense and reconstruct signals/images with lower sampling rate than Nyquist rate. Applications in resource constrained environment stand to benefit from this theory, opening up many possibilities for new applications at the same time. The traditional inference pipeline for computer vision sequence reconstructing the image from compressive measurements. However,the reconstruction process is a computationally expensive step that also provides poor results at high compression rate. There have been several successful attempts to perform inference tasks directly on compressive measurements such as activity recognition. In this thesis, I am interested to tackle a more challenging vision problem - Visual question answering (VQA) without reconstructing the compressive images. I investigate the feasibility of this problem with a series of experiments, and I evaluate proposed methods on a VQA dataset and discuss promising results and direction for future work.
ContributorsHuang, Li-Chin (Author) / Turaga, Pavan (Thesis advisor) / Yang, Yezhou (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2017
Description
Light field imaging is limited in its computational processing demands of high
sampling for both spatial and angular dimensions. Single-shot light field cameras
sacrifice spatial resolution to sample angular viewpoints, typically by multiplexing
incoming rays onto a 2D sensor array. While this resolution can be recovered using
compressive sensing, these iterative solutions are slow in processing a light field. We
present a deep learning approach using a new, two branch network architecture,
consisting jointly of an autoencoder and a 4D CNN, to recover a high resolution
4D light field from a single coded 2D image. This network decreases reconstruction
time significantly while achieving average PSNR values of 26-32 dB on a variety of
light fields. In particular, reconstruction time is decreased from 35 minutes to 6.7
minutes as compared to the dictionary method for equivalent visual quality. These
reconstructions are performed at small sampling/compression ratios as low as 8%,
allowing for cheaper coded light field cameras. We test our network reconstructions
on synthetic light fields, simulated coded measurements of real light fields captured
from a Lytro Illum camera, and real coded images from a custom CMOS diffractive
light field camera. The combination of compressive light field capture with deep
learning allows the potential for real-time light field video acquisition systems in the
future.
sampling for both spatial and angular dimensions. Single-shot light field cameras
sacrifice spatial resolution to sample angular viewpoints, typically by multiplexing
incoming rays onto a 2D sensor array. While this resolution can be recovered using
compressive sensing, these iterative solutions are slow in processing a light field. We
present a deep learning approach using a new, two branch network architecture,
consisting jointly of an autoencoder and a 4D CNN, to recover a high resolution
4D light field from a single coded 2D image. This network decreases reconstruction
time significantly while achieving average PSNR values of 26-32 dB on a variety of
light fields. In particular, reconstruction time is decreased from 35 minutes to 6.7
minutes as compared to the dictionary method for equivalent visual quality. These
reconstructions are performed at small sampling/compression ratios as low as 8%,
allowing for cheaper coded light field cameras. We test our network reconstructions
on synthetic light fields, simulated coded measurements of real light fields captured
from a Lytro Illum camera, and real coded images from a custom CMOS diffractive
light field camera. The combination of compressive light field capture with deep
learning allows the potential for real-time light field video acquisition systems in the
future.
ContributorsGupta, Mayank (Author) / Turaga, Pavan (Thesis advisor) / Yang, Yezhou (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2017
Description
When dancers are granted agency over music, as in interactive dance systems, the actors are most often concerned with the problem of creating a staged performance for an audience. However, as is reflected by the above quote, the practice of Argentine tango social dance is most concerned with participants internal experience and their relationship to the broader tango community. In this dissertation I explore creative approaches to enrich the sense of connection, that is, the experience of oneness with a partner and complete immersion in music and dance for Argentine tango dancers by providing agency over musical activities through the use of interactive technology. Specifically, I create an interactive dance system that allows tango dancers to affect and create music via their movements in the context of social dance. The motivations for this work are multifold: 1) to intensify embodied experience of the interplay between dance and music, individual and partner, couple and community, 2) to create shared experience of the conventions of tango dance, and 3) to innovate Argentine tango social dance practice for the purposes of education and increasing musicality in dancers.
ContributorsBrown, Courtney Douglass (Author) / Paine, Garth (Thesis advisor) / Feisst, Sabine (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2017