Matching Items (14)
Filtering by
- All Subjects: deep learning
- All Subjects: Signal Processing
Description
Detecting early signs of neurodegeneration is vital for measuring the efficacy of pharmaceuticals and planning treatments for neurological diseases. This is especially true for Amyotrophic Lateral Sclerosis (ALS) where differences in symptom onset can be indicative of the prognosis. Because it can be measured noninvasively, changes in speech production have been proposed as a promising indicator of neurological decline. However, speech changes are typically measured subjectively by a clinician. These perceptual ratings can vary widely between clinicians and within the same clinician on different patient visits, making clinical ratings less sensitive to subtle early indicators. In this paper, we propose an algorithm for the objective measurement of flutter, a quasi-sinusoidal modulation of fundamental frequency that manifests in the speech of some ALS patients. The algorithm detailed in this paper employs long-term average spectral analysis on the residual F0 track of a sustained phonation to detect the presence of flutter and is robust to longitudinal drifts in F0. The algorithm is evaluated on a longitudinal speech dataset of ALS patients at varying stages in their prognosis. Benchmarking with two stages of perceptual ratings provided by an expert speech pathologist indicate that the algorithm follows perceptual ratings with moderate accuracy and can objectively detect flutter in instances where the variability of the perceptual rating causes uncertainty.
ContributorsPeplinski, Jacob Scott (Author) / Berisha, Visar (Thesis director) / Liss, Julie (Committee member) / Electrical Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2018-05
Description
Dealing with relational data structures is central to a wide-range of applications including social networks, epidemic modeling, molecular chemistry, medicine, energy distribution, and transportation. Machine learning models that can exploit the inherent structural/relational bias in the graph structured data have gained prominence in recent times. A recurring idea that appears in all approaches is to encode the nodes in the graph (or the entire graph) as low-dimensional vectors also known as embeddings, prior to carrying out downstream task-specific learning. It is crucial to eliminate hand-crafted features and instead directly incorporate the structural inductive bias into the deep learning architectures.
In this dissertation, deep learning models that directly operate on graph structured data are proposed for effective representation learning. A literature review on existing graph representation learning is provided in the beginning of the dissertation. The primary focus of dissertation is on building novel graph neural network architectures that are robust against adversarial attacks. The proposed graph neural network models are extended to multiplex graphs (heterogeneous graphs). Finally, a relational neural network model is proposed to operate on a human structural connectome. For every research contribution of this dissertation, several empirical studies are conducted on benchmark datasets. The proposed graph neural network models, approaches, and architectures demonstrate significant performance improvements in comparison to the existing state-of-the-art graph embedding strategies.
ContributorsShanthamallu, Uday Shankar (Author) / Spanias, Andreas (Thesis advisor) / Thiagarajan, Jayaraman J (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Berisha, Visar (Committee member) / Arizona State University (Publisher)
Created2021
Description
Speech analysis for clinical applications has emerged as a burgeoning field, providing valuable insights into an individual's physical and physiological state. Researchers have explored speech features for clinical applications, such as diagnosing, predicting, and monitoring various pathologies. Before presenting the new deep learning frameworks, this thesis introduces a study on conventional acoustic feature changes in subjects with post-traumatic headache (PTH) attributed to mild traumatic brain injury (mTBI). This work demonstrates the effectiveness of using speech signals to assess the pathological status of individuals. At the same time, it highlights some of the limitations of conventional acoustic and linguistic features, such as low repeatability and generalizability. Two critical characteristics of speech features are (1) good robustness, as speech features need to generalize across different corpora, and (2) high repeatability, as speech features need to be invariant to all confounding factors except the pathological state of targets. This thesis presents two research thrusts in the context of speech signals in clinical applications that focus on improving the robustness and repeatability of speech features, respectively. The first thrust introduces a deep learning framework to generate acoustic feature embeddings sensitive to vocal quality and robust across different corpora. A contrastive loss combined with a classification loss is used to train the model jointly, and data-warping techniques are employed to improve the robustness of embeddings. Empirical results demonstrate that the proposed method achieves high in-corpus and cross-corpus classification accuracy and generates good embeddings sensitive to voice quality and robust across different corpora. The second thrust introduces using the intra-class correlation coefficient (ICC) to evaluate the repeatability of embeddings. A novel regularizer, the ICC regularizer, is proposed to regularize deep neural networks to produce embeddings with higher repeatability. This ICC regularizer is implemented and applied to three speech applications: a clinical application, speaker verification, and voice style conversion. The experimental results reveal that the ICC regularizer improves the repeatability of learned embeddings compared to the contrastive loss, leading to enhanced performance in downstream tasks.
ContributorsZhang, Jianwei (Author) / Jayasuriya, Suren (Thesis advisor) / Berisha, Visar (Thesis advisor) / Liss, Julie (Committee member) / Spanias, Andreas (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
Modern machine learning systems leverage data and features from multiple modalities to gain more predictive power. In most scenarios, the modalities are vastly different and the acquired data are heterogeneous in nature. Consequently, building highly effective fusion algorithms is at the core to achieve improved model robustness and inferencing performance. This dissertation focuses on the representation learning approaches as the fusion strategy. Specifically, the objective is to learn the shared latent representation which jointly exploit the structural information encoded in all modalities, such that a straightforward learning model can be adopted to obtain the prediction.
We first consider sensor fusion, a typical multimodal fusion problem critical to building a pervasive computing platform. A systematic fusion technique is described to support both multiple sensors and descriptors for activity recognition. Targeted to learn the optimal combination of kernels, Multiple Kernel Learning (MKL) algorithms have been successfully applied to numerous fusion problems in computer vision etc. Utilizing the MKL formulation, next we describe an auto-context algorithm for learning image context via the fusion with low-level descriptors. Furthermore, a principled fusion algorithm using deep learning to optimize kernel machines is developed. By bridging deep architectures with kernel optimization, this approach leverages the benefits of both paradigms and is applied to a wide variety of fusion problems.
In many real-world applications, the modalities exhibit highly specific data structures, such as time sequences and graphs, and consequently, special design of the learning architecture is needed. In order to improve the temporal modeling for multivariate sequences, we developed two architectures centered around attention models. A novel clinical time series analysis model is proposed for several critical problems in healthcare. Another model coupled with triplet ranking loss as metric learning framework is described to better solve speaker diarization. Compared to state-of-the-art recurrent networks, these attention-based multivariate analysis tools achieve improved performance while having a lower computational complexity. Finally, in order to perform community detection on multilayer graphs, a fusion algorithm is described to derive node embedding from word embedding techniques and also exploit the complementary relational information contained in each layer of the graph.
We first consider sensor fusion, a typical multimodal fusion problem critical to building a pervasive computing platform. A systematic fusion technique is described to support both multiple sensors and descriptors for activity recognition. Targeted to learn the optimal combination of kernels, Multiple Kernel Learning (MKL) algorithms have been successfully applied to numerous fusion problems in computer vision etc. Utilizing the MKL formulation, next we describe an auto-context algorithm for learning image context via the fusion with low-level descriptors. Furthermore, a principled fusion algorithm using deep learning to optimize kernel machines is developed. By bridging deep architectures with kernel optimization, this approach leverages the benefits of both paradigms and is applied to a wide variety of fusion problems.
In many real-world applications, the modalities exhibit highly specific data structures, such as time sequences and graphs, and consequently, special design of the learning architecture is needed. In order to improve the temporal modeling for multivariate sequences, we developed two architectures centered around attention models. A novel clinical time series analysis model is proposed for several critical problems in healthcare. Another model coupled with triplet ranking loss as metric learning framework is described to better solve speaker diarization. Compared to state-of-the-art recurrent networks, these attention-based multivariate analysis tools achieve improved performance while having a lower computational complexity. Finally, in order to perform community detection on multilayer graphs, a fusion algorithm is described to derive node embedding from word embedding techniques and also exploit the complementary relational information contained in each layer of the graph.
ContributorsSong, Huan (Author) / Spanias, Andreas (Thesis advisor) / Thiagarajan, Jayaraman (Committee member) / Berisha, Visar (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Arizona State University (Publisher)
Created2018
Description
Machine learning (ML) has played an important role in several modern technological innovations and has become an important tool for researchers in various fields of interest. Besides engineering, ML techniques have started to spread across various departments of study, like health-care, medicine, diagnostics, social science, finance, economics etc. These techniques require data to train the algorithms and model a complex system and make predictions based on that model. Due to development of sophisticated sensors it has become easier to collect large volumes of data which is used to make necessary hypotheses using ML. The promising results obtained using ML have opened up new opportunities of research across various departments and this dissertation is a manifestation of it. Here, some unique studies have been presented, from which valuable inference have been drawn for a real-world complex system. Each study has its own unique sets of motivation and relevance to the real world. An ensemble of signal processing (SP) and ML techniques have been explored in each study. This dissertation provides the detailed systematic approach and discusses the results achieved in each study. Valuable inferences drawn from each study play a vital role in areas of science and technology, and it is worth further investigation. This dissertation also provides a set of useful SP and ML tools for researchers in various fields of interest.
ContributorsDutta, Arindam (Author) / Bliss, Daniel W (Thesis advisor) / Berisha, Visar (Committee member) / Richmond, Christ (Committee member) / Corman, Steven (Committee member) / Arizona State University (Publisher)
Created2018
Description
Deep neural networks (DNN) have shown tremendous success in various cognitive tasks, such as image classification, speech recognition, etc. However, their usage on resource-constrained edge devices has been limited due to high computation and large memory requirement.
To overcome these challenges, recent works have extensively investigated model compression techniques such as element-wise sparsity, structured sparsity and quantization. While most of these works have applied these compression techniques in isolation, there have been very few studies on application of quantization and structured sparsity together on a DNN model.
This thesis co-optimizes structured sparsity and quantization constraints on DNN models during training. Specifically, it obtains optimal setting of 2-bit weight and 2-bit activation coupled with 4X structured compression by performing combined exploration of quantization and structured compression settings. The optimal DNN model achieves 50X weight memory reduction compared to floating-point uncompressed DNN. This memory saving is significant since applying only structured sparsity constraints achieves 2X memory savings and only quantization constraints achieves 16X memory savings. The algorithm has been validated on both high and low capacity DNNs and on wide-sparse and deep-sparse DNN models. Experiments demonstrated that deep-sparse DNN outperforms shallow-dense DNN with varying level of memory savings depending on DNN precision and sparsity levels. This work further proposed a Pareto-optimal approach to systematically extract optimal DNN models from a huge set of sparse and dense DNN models. The resulting 11 optimal designs were further evaluated by considering overall DNN memory which includes activation memory and weight memory. It was found that there is only a small change in the memory footprint of the optimal designs corresponding to the low sparsity DNNs. However, activation memory cannot be ignored for high sparsity DNNs.
To overcome these challenges, recent works have extensively investigated model compression techniques such as element-wise sparsity, structured sparsity and quantization. While most of these works have applied these compression techniques in isolation, there have been very few studies on application of quantization and structured sparsity together on a DNN model.
This thesis co-optimizes structured sparsity and quantization constraints on DNN models during training. Specifically, it obtains optimal setting of 2-bit weight and 2-bit activation coupled with 4X structured compression by performing combined exploration of quantization and structured compression settings. The optimal DNN model achieves 50X weight memory reduction compared to floating-point uncompressed DNN. This memory saving is significant since applying only structured sparsity constraints achieves 2X memory savings and only quantization constraints achieves 16X memory savings. The algorithm has been validated on both high and low capacity DNNs and on wide-sparse and deep-sparse DNN models. Experiments demonstrated that deep-sparse DNN outperforms shallow-dense DNN with varying level of memory savings depending on DNN precision and sparsity levels. This work further proposed a Pareto-optimal approach to systematically extract optimal DNN models from a huge set of sparse and dense DNN models. The resulting 11 optimal designs were further evaluated by considering overall DNN memory which includes activation memory and weight memory. It was found that there is only a small change in the memory footprint of the optimal designs corresponding to the low sparsity DNNs. However, activation memory cannot be ignored for high sparsity DNNs.
ContributorsSrivastava, Gaurav (Author) / Seo, Jae-Sun (Thesis advisor) / Chakrabarti, Chaitali (Committee member) / Berisha, Visar (Committee member) / Arizona State University (Publisher)
Created2018
Description
In recent years, conventional convolutional neural network (CNN) has achieved outstanding performance in image and speech processing applications. Unfortunately, the pooling operation in CNN ignores important spatial information which is an important attribute in many applications. The recently proposed capsule network retains spatial information and improves the capabilities of traditional CNN. It uses capsules to describe features in multiple dimensions and dynamic routing to increase the statistical stability of the network.
In this work, we first use capsule network for overlapping digit recognition problem. We evaluate the performance of the network with respect to recognition accuracy, convergence and training time per epoch. We show that capsule network achieves higher accuracy when training set size is small. When training set size is larger, capsule network and conventional CNN have comparable recognition accuracy. The training time per epoch for capsule network is longer than conventional CNN because of the dynamic routing algorithm. An analysis of the GPU timing shows that adjusting the capsule structure can help decrease the time complexity of the dynamic routing algorithm significantly.
Next, we design a capsule network for speech recognition, specifically, overlapping word recognition. We use both capsule network and conventional CNN to recognize 2 overlapping words in speech files created from 5 word classes. We show that capsule network achieves a considerably higher recognition accuracy (96.92%) compared to conventional CNN (85.19%). Our results show that capsule network recognizes overlapping word by recognizing each individual word in the speech. We also verify the scalability of capsule network by increasing the number of word classes from 5 to 10. Capsule network still shows a high recognition accuracy of 95.42% in case of 10 words while the accuracy of conventional CNN decreases sharply to 73.18%.
In this work, we first use capsule network for overlapping digit recognition problem. We evaluate the performance of the network with respect to recognition accuracy, convergence and training time per epoch. We show that capsule network achieves higher accuracy when training set size is small. When training set size is larger, capsule network and conventional CNN have comparable recognition accuracy. The training time per epoch for capsule network is longer than conventional CNN because of the dynamic routing algorithm. An analysis of the GPU timing shows that adjusting the capsule structure can help decrease the time complexity of the dynamic routing algorithm significantly.
Next, we design a capsule network for speech recognition, specifically, overlapping word recognition. We use both capsule network and conventional CNN to recognize 2 overlapping words in speech files created from 5 word classes. We show that capsule network achieves a considerably higher recognition accuracy (96.92%) compared to conventional CNN (85.19%). Our results show that capsule network recognizes overlapping word by recognizing each individual word in the speech. We also verify the scalability of capsule network by increasing the number of word classes from 5 to 10. Capsule network still shows a high recognition accuracy of 95.42% in case of 10 words while the accuracy of conventional CNN decreases sharply to 73.18%.
ContributorsXiong, Yan (Author) / Chakrabarti, Chaitali (Thesis advisor) / Berisha, Visar (Thesis advisor) / Weng, Yang (Committee member) / Arizona State University (Publisher)
Created2018
Description
In recent years, there has been an increased interest in sharing available bandwidth to avoid spectrum congestion. With an ever-increasing number wireless users, it is critical to develop signal processing based spectrum sharing algorithms to achieve cooperative use of the allocated spectrum among multiple systems in order to reduce interference between systems. This work studies the radar and communications systems coexistence problem using two main approaches. The first approach develops methodologies to increase radar target tracking performance under low signal-to-interference-plus-noise ratio (SINR) conditions due to the coexistence of strong communications interference. The second approach jointly optimizes the performance of both systems by co-designing a common transmit waveform.
When concentrating on improving radar tracking performance, a pulsed radar that is tracking a single target coexisting with high powered communications interference is considered. Although the Cramer-Rao lower bound (CRLB) on the covariance of an unbiased estimator of deterministic parameters provides a bound on the estimation mean squared error (MSE), there exists an SINR threshold at which estimator covariance rapidly deviates from the CRLB. After demonstrating that different radar waveforms experience different estimation SINR thresholds using the Barankin bound (BB), a new radar waveform design method is proposed based on predicting the waveform-dependent BB SINR threshold under low SINR operating conditions.
A novel method of predicting the SINR threshold value for maximum likelihood estimation (MLE) is proposed. A relationship is shown to exist between the formulation of the BB kernel and the probability of selecting sidelobes for the MLE. This relationship is demonstrated as an accurate means of threshold prediction for the radar target parameter estimation of frequency, time-delay and angle-of-arrival.
For the co-design radar and communications system problem, the use of a common transmit waveform for a pulse-Doppler radar and a multiuser communications system is proposed. The signaling scheme for each system is selected from a class of waveforms with nonlinear phase function by optimizing the waveform parameters to minimize interference between the two systems and interference among communications users. Using multi-objective optimization, a trade-off in system performance is demonstrated when selecting waveforms that minimize both system interference and tracking MSE.
When concentrating on improving radar tracking performance, a pulsed radar that is tracking a single target coexisting with high powered communications interference is considered. Although the Cramer-Rao lower bound (CRLB) on the covariance of an unbiased estimator of deterministic parameters provides a bound on the estimation mean squared error (MSE), there exists an SINR threshold at which estimator covariance rapidly deviates from the CRLB. After demonstrating that different radar waveforms experience different estimation SINR thresholds using the Barankin bound (BB), a new radar waveform design method is proposed based on predicting the waveform-dependent BB SINR threshold under low SINR operating conditions.
A novel method of predicting the SINR threshold value for maximum likelihood estimation (MLE) is proposed. A relationship is shown to exist between the formulation of the BB kernel and the probability of selecting sidelobes for the MLE. This relationship is demonstrated as an accurate means of threshold prediction for the radar target parameter estimation of frequency, time-delay and angle-of-arrival.
For the co-design radar and communications system problem, the use of a common transmit waveform for a pulse-Doppler radar and a multiuser communications system is proposed. The signaling scheme for each system is selected from a class of waveforms with nonlinear phase function by optimizing the waveform parameters to minimize interference between the two systems and interference among communications users. Using multi-objective optimization, a trade-off in system performance is demonstrated when selecting waveforms that minimize both system interference and tracking MSE.
ContributorsKota, John S (Author) / Papandreou-Suppappola, Antonia (Thesis advisor) / Berisha, Visar (Committee member) / Bliss, Daniel (Committee member) / Kovvali, Narayan (Committee member) / Arizona State University (Publisher)
Created2016
Description
This work considers the problem of multiple detection and tracking in two complex time-varying environments, urban terrain and underwater. Tracking multiple radar targets in urban environments is rst investigated by exploiting multipath signal returns, wideband underwater acoustic (UWA) communications channels are estimated using adaptive learning methods, and multiple UWA communications users are detected by designing the transmit signal to match the environment. For the urban environment, a multi-target tracking algorithm is proposed that integrates multipath-to-measurement association and the probability hypothesis density method implemented using particle filtering. The algorithm is designed to track an unknown time-varying number of targets by extracting information from multiple measurements due to multipath returns in the urban terrain. The path likelihood probability is calculated by considering associations between measurements and multipath returns, and an adaptive clustering algorithm is used to estimate the number of target and their corresponding parameters. The performance of the proposed algorithm is demonstrated for different multiple target scenarios and evaluated using the optimal subpattern assignment metric. The underwater environment provides a very challenging communication channel due to its highly time-varying nature, resulting in large distortions due to multipath and Doppler-scaling, and frequency-dependent path loss. A model-based wideband UWA channel estimation algorithm is first proposed to estimate the channel support and the wideband spreading function coefficients. A nonlinear frequency modulated signaling scheme is proposed that is matched to the wideband characteristics of the underwater environment. Constraints on the signal parameters are derived to optimally reduce multiple access interference and the UWA channel effects. The signaling scheme is compared to a code division multiple access (CDMA) scheme to demonstrate its improved bit error rate performance. The overall multi-user communication system performance is finally analyzed by first estimating the UWA channel and then designing the signaling scheme for multiple communications users.
ContributorsZhou, Meng (Author) / Papandreou-Suppappola, Antonia (Thesis advisor) / Tepedelenlioğlu, Cihan (Committee member) / Kovvali, Narayan (Committee member) / Berisha, Visar (Committee member) / Arizona State University (Publisher)
Created2014