Matching Items (10)
Filtering by
- All Subjects: Coding theory
- All Subjects: Machine Learning
- Genre: Doctoral Dissertation
- Creators: Tepedelenlioğlu, Cihan
Description
Image understanding has been playing an increasingly crucial role in vision applications. Sparse models form an important component in image understanding, since the statistics of natural images reveal the presence of sparse structure. Sparse methods lead to parsimonious models, in addition to being efficient for large scale learning. In sparse modeling, data is represented as a sparse linear combination of atoms from a "dictionary" matrix. This dissertation focuses on understanding different aspects of sparse learning, thereby enhancing the use of sparse methods by incorporating tools from machine learning. With the growing need to adapt models for large scale data, it is important to design dictionaries that can model the entire data space and not just the samples considered. By exploiting the relation of dictionary learning to 1-D subspace clustering, a multilevel dictionary learning algorithm is developed, and it is shown to outperform conventional sparse models in compressed recovery, and image denoising. Theoretical aspects of learning such as algorithmic stability and generalization are considered, and ensemble learning is incorporated for effective large scale learning. In addition to building strategies for efficiently implementing 1-D subspace clustering, a discriminative clustering approach is designed to estimate the unknown mixing process in blind source separation. By exploiting the non-linear relation between the image descriptors, and allowing the use of multiple features, sparse methods can be made more effective in recognition problems. The idea of multiple kernel sparse representations is developed, and algorithms for learning dictionaries in the feature space are presented. Using object recognition experiments on standard datasets it is shown that the proposed approaches outperform other sparse coding-based recognition frameworks. Furthermore, a segmentation technique based on multiple kernel sparse representations is developed, and successfully applied for automated brain tumor identification. Using sparse codes to define the relation between data samples can lead to a more robust graph embedding for unsupervised clustering. By performing discriminative embedding using sparse coding-based graphs, an algorithm for measuring the glomerular number in kidney MRI images is developed. Finally, approaches to build dictionaries for local sparse coding of image descriptors are presented, and applied to object recognition and image retrieval.
ContributorsJayaraman Thiagarajan, Jayaraman (Author) / Spanias, Andreas (Thesis advisor) / Frakes, David (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2013
Description
There are many wireless communication and networking applications that require high transmission rates and reliability with only limited resources in terms of bandwidth, power, hardware complexity etc.. Real-time video streaming, gaming and social networking are a few such examples. Over the years many problems have been addressed towards the goal of enabling such applications; however, significant challenges still remain, particularly, in the context of multi-user communications. With the motivation of addressing some of these challenges, the main focus of this dissertation is the design and analysis of capacity approaching coding schemes for several (wireless) multi-user communication scenarios. Specifically, three main themes are studied: superposition coding over broadcast channels, practical coding for binary-input binary-output broadcast channels, and signalling schemes for two-way relay channels. As the first contribution, we propose an analytical tool that allows for reliable comparison of different practical codes and decoding strategies over degraded broadcast channels, even for very low error rates for which simulations are impractical. The second contribution deals with binary-input binary-output degraded broadcast channels, for which an optimal encoding scheme that achieves the capacity boundary is found, and a practical coding scheme is given by concatenation of an outer low density parity check code and an inner (non-linear) mapper that induces desired distribution of "one" in a codeword. The third contribution considers two-way relay channels where the information exchange between two nodes takes place in two transmission phases using a coding scheme called physical-layer network coding. At the relay, a near optimal decoding strategy is derived using a list decoding algorithm, and an approximation is obtained by a joint decoding approach. For the latter scheme, an analytical approximation of the word error rate based on a union bounding technique is computed under the assumption that linear codes are employed at the two nodes exchanging data. Further, when the wireless channel is frequency selective, two decoding strategies at the relay are developed, namely, a near optimal decoding scheme implemented using list decoding, and a reduced complexity detection/decoding scheme utilizing a linear minimum mean squared error based detector followed by a network coded sequence decoder.
ContributorsBhat, Uttam (Author) / Duman, Tolga M. (Thesis advisor) / Tepedelenlioğlu, Cihan (Committee member) / Li, Baoxin (Committee member) / Zhang, Junshan (Committee member) / Arizona State University (Publisher)
Created2011
Description
Following the success in incorporating perceptual models in audio coding algorithms, their application in other speech/audio processing systems is expanding. In general, all perceptual speech/audio processing algorithms involve minimization of an objective function that directly/indirectly incorporates properties of human perception. This dissertation primarily investigates the problems associated with directly embedding an auditory model in the objective function formulation and proposes possible solutions to overcome high complexity issues for use in real-time speech/audio algorithms. Specific problems addressed in this dissertation include: 1) the development of approximate but computationally efficient auditory model implementations that are consistent with the principles of psychoacoustics, 2) the development of a mapping scheme that allows synthesizing a time/frequency domain representation from its equivalent auditory model output. The first problem is aimed at addressing the high computational complexity involved in solving perceptual objective functions that require repeated application of auditory model for evaluation of different candidate solutions. In this dissertation, a frequency pruning and a detector pruning algorithm is developed that efficiently implements the various auditory model stages. The performance of the pruned model is compared to that of the original auditory model for different types of test signals in the SQAM database. Experimental results indicate only a 4-7% relative error in loudness while attaining up to 80-90 % reduction in computational complexity. Similarly, a hybrid algorithm is developed specifically for use with sinusoidal signals and employs the proposed auditory pattern combining technique together with a look-up table to store representative auditory patterns. The second problem obtains an estimate of the auditory representation that minimizes a perceptual objective function and transforms the auditory pattern back to its equivalent time/frequency representation. This avoids the repeated application of auditory model stages to test different candidate time/frequency vectors in minimizing perceptual objective functions. In this dissertation, a constrained mapping scheme is developed by linearizing certain auditory model stages that ensures obtaining a time/frequency mapping corresponding to the estimated auditory representation. This paradigm was successfully incorporated in a perceptual speech enhancement algorithm and a sinusoidal component selection task.
ContributorsKrishnamoorthi, Harish (Author) / Spanias, Andreas (Thesis advisor) / Papandreou-Suppappola, Antonia (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Tsakalis, Konstantinos (Committee member) / Arizona State University (Publisher)
Created2011
Description
Insertion and deletion errors represent an important category of channel impairments. Despite their importance and much work over the years, channels with such impairments are far from being fully understood as they proved to be difficult to analyze. In this dissertation, a promising coding scheme is investigated over independent and identically distributed (i.i.d.) insertion/deletion channels, i.e., interleaved concatenation of an outer low-density parity-check (LDPC) code with error-correction capabilities and an inner marker code for synchronization purposes. Marker code structures which offer the highest achievable rates are found with standard bit-level synchronization is performed. Then, to exploit the correlations in the likelihoods corresponding to different transmitted bits, a novel symbol-level synchronization algorithm that works on groups of consecutive bits is introduced. Extrinsic information transfer (EXIT) charts are also utilized to analyze the convergence behavior of the receiver, and to design LDPC codes with degree distributions matched to these channels. The next focus is on segmented deletion channels. It is first shown that such channels are information stable, and hence their channel capacity exists. Several upper and lower bounds are then introduced in an attempt to understand the channel capacity behavior. The asymptotic behavior of the channel capacity is also quantified when the average bit deletion rate is small. Further, maximum-a-posteriori (MAP) based synchronization algorithms are developed and specific LDPC codes are designed to match the channel characteristics. Finally, in addition to binary substitution errors, coding schemes and the corresponding detection algorithms are also studied for several other models with synchronization errors, including inter-symbol interference (ISI) channels, channels with multiple transmit/receive elements and multi-user communication systems.
ContributorsWang, Feng (Author) / Duman, Tolga M. (Thesis advisor) / Tepedelenlioğlu, Cihan (Committee member) / Reisslein, Martin (Committee member) / Zhang, Junshan (Committee member) / Arizona State University (Publisher)
Created2012
Description
The presence of strategic agents can pose unique challenges to data collection and distributed learning. This dissertation first explores the social network dimension of data collection markets, and then focuses on how the strategic agents can be efficiently and effectively incentivized to cooperate in distributed machine learning frameworks.
The first problem explores the impact of social learning in collecting and trading unverifiable information where a data collector purchases data from users through a payment mechanism. Each user starts with a personal signal which represents the knowledge about the underlying state the data collector desires to learn. Through social interactions, each user also acquires additional information from his neighbors in the social network. It is revealed that both the data collector and the users can benefit from social learning which drives down the privacy costs and helps to improve the state estimation for a given total payment budget. In the second half, a federated learning scheme to train a global learning model with strategic agents, who are not bound to contribute their resources unconditionally, is considered. Since the agents are not obliged to provide their true stochastic gradient updates and the server is not capable of directly validating the authenticity of reported updates, the learning process may reach a noncooperative equilibrium. First, the actions of the agents are assumed to be binary: cooperative or defective. If the cooperative action is taken, the agent sends a privacy-preserved version of stochastic gradient signal. If the defective action is taken, the agent sends an arbitrary uninformative noise signal. Furthermore, this setup is extended into the scenarios with more general actions spaces where the quality of the stochastic gradient updates have a range of discrete levels. The proposed methodology evaluates each agent's stochastic gradient according to a reference gradient estimate which is constructed from the gradients provided by other agents, and rewards the agent based on that evaluation.
ContributorsAkbay, Abdullah Basar (Author) / Tepedelenlioğlu, Cihan (Thesis advisor) / Spanias, Andreas (Committee member) / Kosut, Oliver (Committee member) / Ewaisha, Ahmed (Committee member) / Arizona State University (Publisher)
Created2023
Description
The past decade witnessed the success of deep learning models in various applications of computer vision and natural language processing. This success can be predominantly attributed to the (i) availability of large amounts of training data; (ii) access of domain aware knowledge; (iii) i.i.d assumption between the train and target distributions and (iv) belief on existing metrics as reliable indicators of performance. When any of these assumptions are violated, the models exhibit brittleness producing adversely varied behavior. This dissertation focuses on methods for accurate model design and characterization that enhance process reliability when certain assumptions are not met. With the need to safely adopt artificial intelligence tools in practice, it is vital to build reliable failure detectors that indicate regimes where the model must not be invoked. To that end, an error predictor trained with a self-calibration objective is developed to estimate loss consistent with the underlying model. The properties of the error predictor are described and their utility in supporting introspection via feature importances and counterfactual explanations is elucidated. While such an approach can signal data regime changes, it is critical to calibrate models using regimes of inlier (training) and outlier data to prevent under- and over-generalization in models i.e., incorrectly identifying inliers as outliers and vice-versa. By identifying the space for specifying inliers and outliers, an anomaly detector that can effectively flag data of varying semantic complexities in medical imaging is next developed. Uncertainty quantification in deep learning models involves identifying sources of failure and characterizing model confidence to enable actionability. A training strategy is developed that allows the accurate estimation of model uncertainties and its benefits are demonstrated for active learning and generalization gap prediction. This helps identify insufficiently sampled regimes and representation insufficiency in models. In addition, the task of deep inversion under data scarce scenarios is considered, which in practice requires a prior to control the optimization. By identifying limitations in existing work, data priors powered by generative models and deep model priors are designed for audio restoration. With relevant empirical studies on a variety of benchmarks, the need for such design strategies is demonstrated.
ContributorsNarayanaswamy, Vivek Sivaraman (Author) / Spanias, Andreas (Thesis advisor) / J. Thiagarajan, Jayaraman (Committee member) / Berisha, Visar (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Arizona State University (Publisher)
Created2023
Description
There has been a lot of work on the characterization of capacity and achievable rate regions, and rate region outer-bounds for various multi-user channels of interest. Parallel to the developed information theoretic results, practical codes have also been designed for some multi-user channels such as multiple access channels, broadcast channels and relay channels; however, interference channels have not received much attention and only a limited amount of work has been conducted on them. With this motivation, in this dissertation, design of practical and implementable channel codes is studied focusing on multi-user channels with special emphasis on interference channels; in particular, irregular low-density-parity-check codes are exploited for a variety of cases and trellis based codes for short block length designs are performed.
Novel code design approaches are first studied for the two-user Gaussian multiple access channel. Exploiting Gaussian mixture approximation, new methods are proposed wherein the optimized codes are shown to improve upon the available designs and off-the-shelf point-to-point codes applied to the multiple access channel scenario. The code design is then examined for the two-user Gaussian interference channel implementing the Han-Kobayashi encoding and decoding strategy. Compared with the point-to-point codes, the newly designed codes consistently offer better performance. Parallel to this work, code design is explored for the discrete memoryless interference channels wherein the channel inputs and outputs are taken from a finite alphabet and it is demonstrated that the designed codes are superior to the single user codes used with time sharing. Finally, the code design principles are also investigated for the two-user Gaussian interference channel employing trellis-based codes with short block lengths for the case of strong and mixed interference levels.
Novel code design approaches are first studied for the two-user Gaussian multiple access channel. Exploiting Gaussian mixture approximation, new methods are proposed wherein the optimized codes are shown to improve upon the available designs and off-the-shelf point-to-point codes applied to the multiple access channel scenario. The code design is then examined for the two-user Gaussian interference channel implementing the Han-Kobayashi encoding and decoding strategy. Compared with the point-to-point codes, the newly designed codes consistently offer better performance. Parallel to this work, code design is explored for the discrete memoryless interference channels wherein the channel inputs and outputs are taken from a finite alphabet and it is demonstrated that the designed codes are superior to the single user codes used with time sharing. Finally, the code design principles are also investigated for the two-user Gaussian interference channel employing trellis-based codes with short block lengths for the case of strong and mixed interference levels.
ContributorsSharifi, Shahrouz (Author) / Duman, Tolga M. (Thesis advisor) / Zhang, Junshan (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Reisslein, Martin (Committee member) / Arizona State University (Publisher)
Created2015
Description
The recent proposal of two-way relaying has attracted much attention due to its promising features for many practical scenarios. Hereby, two users communicate simultaneously in both directions to exchange their messages with the help of a relay node. This doctoral study investigates various aspects of two-way relaying. Specifically, the issue of asynchronism, lack of channel knowledge, transmission of correlated sources and multi-way relaying techniques involving multiple users are explored.
With the motivation of developing enabling techniques for two-way relay (TWR) channels experiencing excessive synchronization errors, two conceptually-different schemes are proposed to accommodate any relative misalignment between the signals received at any node. By designing a practical transmission/detection mechanism based on orthogonal frequency division multiplexing (OFDM), the proposed schemes perform significantly better than existing competing solutions. In a related direction, differential modulation is implemented for asynchronous TWR systems that lack the channel state information (CSI) knowledge. The challenge in this problem compared to the conventional point-to-point counterpart arises not only from the asynchrony but also from the existence of an interfering signal. Extensive numerical examples, supported by analytical work, are given to demonstrate the advantages of the proposed schemes.
Other important issues considered in this dissertation are related to the extension of the two-way relaying scheme to the multiple-user case, known as the multi-way relaying. First, a distributed source coding solution based on Slepian-Wolf coding is proposed to compress correlated messages close to the information theoretical limits in the context of multi-way relay (MWR) channels. Specifically, the syndrome approach based on low-density parity-check (LDPC) codes is implemented. A number of relaying strategies are considered for this problem offering a tradeoff between performance and complexity. The proposed solutions have shown significant improvements compared to the existing ones in terms of the achievable compression rates. On a different front, a novel approach to channel coding is proposed for the MWR channel based on the implementation of nested codes in a distributed manner. This approach ensures that each node decodes the messages of the other users without requiring complex operations at the relay, and at the same time, providing substantial benefits compared to the traditional routing solution.
With the motivation of developing enabling techniques for two-way relay (TWR) channels experiencing excessive synchronization errors, two conceptually-different schemes are proposed to accommodate any relative misalignment between the signals received at any node. By designing a practical transmission/detection mechanism based on orthogonal frequency division multiplexing (OFDM), the proposed schemes perform significantly better than existing competing solutions. In a related direction, differential modulation is implemented for asynchronous TWR systems that lack the channel state information (CSI) knowledge. The challenge in this problem compared to the conventional point-to-point counterpart arises not only from the asynchrony but also from the existence of an interfering signal. Extensive numerical examples, supported by analytical work, are given to demonstrate the advantages of the proposed schemes.
Other important issues considered in this dissertation are related to the extension of the two-way relaying scheme to the multiple-user case, known as the multi-way relaying. First, a distributed source coding solution based on Slepian-Wolf coding is proposed to compress correlated messages close to the information theoretical limits in the context of multi-way relay (MWR) channels. Specifically, the syndrome approach based on low-density parity-check (LDPC) codes is implemented. A number of relaying strategies are considered for this problem offering a tradeoff between performance and complexity. The proposed solutions have shown significant improvements compared to the existing ones in terms of the achievable compression rates. On a different front, a novel approach to channel coding is proposed for the MWR channel based on the implementation of nested codes in a distributed manner. This approach ensures that each node decodes the messages of the other users without requiring complex operations at the relay, and at the same time, providing substantial benefits compared to the traditional routing solution.
ContributorsSalīm, Aḥmad (Author) / Duman, Tolga M. (Thesis advisor) / Papandreou-Suppappola, Antonia (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Zhang, Junshan (Committee member) / Arizona State University (Publisher)
Created2015
Description
In recent years, the widespread use of deep neural networks (DNNs) has facilitated great improvements in performance for computer vision tasks like image classification and object recognition. In most realistic computer vision applications, an input image undergoes some form of image distortion such as blur and additive noise during image acquisition or transmission. Deep networks trained on pristine images perform poorly when tested on such distortions. DNN predictions have also been shown to be vulnerable to carefully crafted adversarial perturbations. Specifically, so-called universal adversarial perturbations are image-agnostic perturbations that can be added to any image and can fool a target network into making erroneous predictions. This work proposes selective DNN feature regeneration to improve the robustness of existing DNNs to image distortions and universal adversarial perturbations.
In the context of common naturally occurring image distortions, a metric is proposed to identify the most susceptible DNN convolutional filters and rank them in order of the highest gain in classification accuracy upon correction. The proposed approach called DeepCorrect applies small stacks of convolutional layers with residual connections at the output of these ranked filters and trains them to correct the most distortion-affected filter activations, whilst leaving the rest of the pre-trained filter outputs in the network unchanged. Performance results show that applying DeepCorrect models for common vision tasks significantly improves the robustness of DNNs against distorted images and outperforms other alternative approaches.
In the context of universal adversarial perturbations, departing from existing defense strategies that work mostly in the image domain, a novel and effective defense which only operates in the DNN feature domain is presented. This approach identifies pre-trained convolutional features that are most vulnerable to adversarial perturbations and deploys trainable feature regeneration units which transform these DNN filter activations into resilient features that are robust to universal perturbations. Regenerating only the top 50% adversarially susceptible activations in at most 6 DNN layers and leaving all remaining DNN activations unchanged can outperform existing defense strategies across different network architectures and across various universal attacks.
In the context of common naturally occurring image distortions, a metric is proposed to identify the most susceptible DNN convolutional filters and rank them in order of the highest gain in classification accuracy upon correction. The proposed approach called DeepCorrect applies small stacks of convolutional layers with residual connections at the output of these ranked filters and trains them to correct the most distortion-affected filter activations, whilst leaving the rest of the pre-trained filter outputs in the network unchanged. Performance results show that applying DeepCorrect models for common vision tasks significantly improves the robustness of DNNs against distorted images and outperforms other alternative approaches.
In the context of universal adversarial perturbations, departing from existing defense strategies that work mostly in the image domain, a novel and effective defense which only operates in the DNN feature domain is presented. This approach identifies pre-trained convolutional features that are most vulnerable to adversarial perturbations and deploys trainable feature regeneration units which transform these DNN filter activations into resilient features that are robust to universal perturbations. Regenerating only the top 50% adversarially susceptible activations in at most 6 DNN layers and leaving all remaining DNN activations unchanged can outperform existing defense strategies across different network architectures and across various universal attacks.
ContributorsBorkar, Tejas Shyam (Author) / Karam, Lina J (Thesis advisor) / Turaga, Pavan (Committee member) / Jayasuriya, Suren (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Arizona State University (Publisher)
Created2020
Description
In many real-world machine learning classification applications, well labeled training data can be difficult, expensive, or even impossible to obtain. In such situations, it is sometimes possible to label a small subset of data as belonging to the class of interest though it is impractical to manually label all data not of interest. The result is a small set of positive labeled data and a large set of unknown and unlabeled data. This is known as the Positive and Unlabeled learning (PU learning) problem, a type of semi-supervised learning. In this dissertation, the PU learning problem is rigorously defined, several common assumptions described, and a literature review of the field provided. A new family of effective PU learning algorithms, the MLR (Modified Logistic Regression) family of algorithms, is described. Theoretical and experimental justification for these algorithms is provided demonstrating their success and flexibility. Extensive experimentation and empirical evidence are provided comparing several new and existing PU learning evaluation estimation metrics in a wide variety of scenarios. The surprisingly clear advantage of a simple recall estimate as the best estimate for overall PU classifier performance is described. Finally, an application of PU learning to the field of solar fault detection, an area not previously explored in the field, demonstrates the advantage and potential of PU learning in new application domains.
ContributorsJaskie, Kristen P (Author) / Spanias, Andreas (Thesis advisor) / Blain-Christen, Jennifer (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Thiagarajan, Jayaraman (Committee member) / Arizona State University (Publisher)
Created2021