Deep Learning Methods For Mining Genomic Sequence Patterns, Xin Gao

Dissertations

Nowadays, with the growing availability of large-scale genomic datasets and advanced computational techniques, more and more data-driven computational methods have been developed to analyze genomic data and help to solve incompletely understood biological problems. Among them, deep learning methods, have been proposed to automatically learn and recognize the functional activity of DNA sequences from genomics data. Techniques for efficient mining genomic sequence pattern will help to improve our understanding of gene regulation, and thus accelerate our progress toward using personal genomes in medicine.

This dissertation focuses on the development of deep learning methods for mining genomic sequences. First, we compare …

Computational Intelligence In Steganography: Adaptive Image Watermarking, Xin Zhong

Dissertations

Digital image watermarking, as an extension of traditional steganography, refers to the process of hiding certain messages into cover images. The transport image, called marked-image or stego-image, conveys the hidden messages while appears visibly similar to the cover-image. Therefore, image watermarking enables various applications such as copyright protection and covert communication. In a watermarking scheme, fidelity, capacity and robustness are considered as crucial factors, where fidelity measures the similarity between the cover- and marked-images, capacity measures the maximum amount of watermark that can be embedded, and robustness concerns the watermark extraction under attacks on the marked-image. Watermarking techniques are often …

Discerning Novel Splice Junctions Derived From Rna-Seq Alignment: A Deep Learning Approach, Yi Zhang, Xinan Liu, James N. Macleod, Jinze Liu

Computer Science Faculty Publications

Background: Exon splicing is a regulated cellular process in the transcription of protein-coding genes. Technological advancements and cost reductions in RNA sequencing have made quantitative and qualitative assessments of the transcriptome both possible and widely available. RNA-seq provides unprecedented resolution to identify gene structures and resolve the diversity of splicing variants. However, currently available ab initio aligners are vulnerable to spurious alignments due to random sequence matches and sample-reference genome discordance. As a consequence, a significant set of false positive exon junction predictions would be introduced, which will further confuse downstream analyses of splice variant discovery and abundance estimation.

Results: …

Efficient Machine Learning: Models And Accelerations, Zhe Li

Dissertations - ALL

One of the key enablers of the recent unprecedented success of machine learning is the adoption of very large models. Modern machine learning models typically consist of multiple cascaded layers such as deep neural networks, and at least millions to hundreds of millions of parameters (i.e., weights) for the entire model. The larger-scale model tend to enable the extraction of more complex high-level features, and therefore, lead to a significant improvement of the overall accuracy. On the other side, the layered deep structure and large model sizes also demand to increase computational capability and memory requirements. In order to achieve …

Artificial Intelligence In Materials Science: Applications Of Machine Learning To Extraction Of Physically Meaningful Information From Atomic Resolution Microscopy Imaging, Artem Borisovich Maksov

Doctoral Dissertations

Materials science is the cornerstone for technological development of the modern world that has been largely shaped by the advances in fabrication of semiconductor materials and devices. However, the Moore’s Law is expected to stop by 2025 due to reaching the limits of traditional transistor scaling. However, the classical approach has shown to be unable to keep up with the needs of materials manufacturing, requiring more than 20 years to move a material from discovery to market. To adapt materials fabrication to the needs of the 21st century, it is necessary to develop methods for much faster processing of experimental …

Retrospective Motion Correction In Magnetic Resonance Imaging Of The Brain, Patricia Johnson

Electronic Thesis and Dissertation Repository

Magnetic Resonance Imaging (MRI) is a tremendously useful diagnostic imaging modality that provides outstanding soft tissue contrast. However, subject motion is a significant unsolved problem; motion during image acquisition can cause blurring and distortions in the image, limiting its diagnostic utility. Current techniques for addressing head motion include optical tracking which can be impractical in clinical settings due to challenges associated with camera cross-calibration and marker fixation. Another category of techniques is MRI navigators, which use specially acquired MRI data to track the motion of the head.

This thesis presents two techniques for motion correction in MRI: the first is …

Objective Assessment Of Machine Learning Algorithms For Speech Enhancement In Hearing Aids, Krishnan Parameswaran

Electronic Thesis and Dissertation Repository

Speech enhancement in assistive hearing devices has been an area of research for many decades. Noise reduction is particularly challenging because of the wide variety of noise sources and the non-stationarity of speech and noise. Digital signal processing (DSP) algorithms deployed in modern hearing aids for noise reduction rely on certain assumptions on the statistical properties of undesired signals. This could be disadvantageous in accurate estimation of different noise types, which subsequently leads to suboptimal noise reduction. In this research, a relatively unexplored technique based on deep learning, i.e. Recurrent Neural Network (RNN), is used to perform noise reduction and …

Multi-Modal Deep Learning To Understand Vision And Language, Shagan Sah

Theses

Developing intelligent agents that can perceive and understand the rich visual world around us has been a long-standing goal in the field of artificial intelligence. In the last few years, significant progress has been made towards this goal and deep learning has been attributed to recent incredible advances in general visual and language understanding. Convolutional neural networks have been used to learn image representations while recurrent neural networks have demonstrated the ability to generate text from visual stimuli. In this thesis, we develop methods and techniques using hybrid convolutional and recurrent neural network architectures that connect visual data and natural …

A Microscopic Simulation Laboratory For Evaluation Of Off-Street Parking Systems, Yun Yuan

Theses and Dissertations

The parking industry produces an enormous amount of data every day that, properly analyzed, will change the way the industry operates. The collected data form patterns that, in most cases, would allow parking operators and property owners to better understand how to maximize revenue and decrease operating expenses and support the decisions such as how to set specific parking policies (e.g. electrical charging only parking space) to achieve the sustainable and eco-friendly parking.

However, there lacks an intelligent tool to assess the layout design and operational performance of parking lots to reduce the externalities and increase the revenue. To address …

Ordinal Hyperplane Loss, Bob Vanderheyden, Ying Xie

OHPL Publications

The problem of ordinal classification occurs in a large and growing number of areas. Some of the most common source and applications of ordinal data include rating scales, medical classification scales, socio-economic scales, meaningful groupings of continuous data, facial emotional intensity, facial age estimation, etc. The problem of predicting ordinal classes is typically addressed by either performing n-1 binary classification for n ordinal classes or treating ordinal classes as continuous values for regression. However, the first strategy doesn’t fully utilize the ordering information of classes and the second strategy imposes a strong continuous assumption to ordinal classes. In this paper, …

Making A Good Thing Better: Enhancing Password/Pin-Based User Authentication With Smartwatch, Bing Chang, Yingjiu Li, Qiongxiao Wang, Wen-Tao Zhu, Robert H. Deng

Research Collection School Of Computing and Information Systems

Wearing smartwatches becomes increasingly popular in people’s lives. This paper shows that a smartwatch can help its bearer authenticate to a login system effectively and securely even if the bearer’s password has already been revealed. This idea is motivated by our observation that a sensor-rich smartwatch is capable of tracking the wrist motions of its bearer typing a password or PIN, which can be used as an authentication factor. The major challenge in this research is that a sophisticated attacker may imitate a user’s typing behavior as shown in previous research on keystroke dynamics based user authentication. We address this …

Towards End-To-End Semi-Supervised Deep Learning For Drug Discovery, Xiaoyu Zhang

Computer Science and Engineering Theses

Observing the recent progress in Deep Learning, the employment of AI is surging to accelerate drug discovery and cut R&D costs in the last few years. However, the success of deep learning is attributed to large-scale clean high-quality labeled data, which is generally unavailable in drug discovery practices. In this thesis, we address this issue by proposing an end-to-end deep learning framework in a semi supervised learning fashion. That is said, the proposed deep learning approach can utilize both labeled and unlabeled data. While labeled data is of very limited availability, the amount of available unlabeled data is generally huge. …

Large-Scale Deep Learning With Application In Medical Imaging And Bio-Informatics, Zheng Xu

Computer Science and Engineering Dissertations

With the recent advancement of the deep learning technology in the artificial intelligence area, nowadays people's lives have been drastically changed. However, the success of deep learning technology mostly relies on large-scale high-quality data-sets. The complexity of deeper model and larger scale datasets have brought us significant challenges. Inspired by this trend, in this dissertation, we focus on developing efficient and effective large-scale deep learning techniques in solving real-world problems, like cell detection in hyper-resolution medical image or drug screening from millions of compound candidates. With respect to the hyper-resolution medical imaging cell detection problem, the challenges are mainly the …

Defending Neural Networks Against Adversarial Examples, Armon Barton

Computer Science and Engineering Dissertations

Deep learning is becoming a technology central to the safety of cars, the security of networks, and the correct functioning of many other types of systems. Unfortunately, attackers can create adversarial examples, small perturbations to inputs that trick deep neural networks into making a misclassification. Researchers have explored various defenses against this attack, but many of them have been broken. The most robust approaches are Adversarial Training and its extension, Adversarial Logit Pairing, but Adversarial Training requires generating and training on adversarial examples from any possible attack. This is not only expensive, but it is inherently vulnerable to novel attack …

Deepsign: A Deep-Learning Architecture For Sign Language, Jai Amrish Shah

Computer Science and Engineering Theses

Sign languages are used by deaf people for communication. In sign languages, humans use hand gestures, body, facial expressions and movements to convey meaning. Humans can easily learn and understand sign languages, but automatic sign language recognition for machines is a challenging task. Using recent advances in the field of deep-learning, we introduce a fully automated deep-learning architecture for isolated sign language recognition. Our architecture tries to address three problems: 1) Satisfactory accuracy with limited data samples 2) Reducing chances of over-fitting when the data is limited 3) Automating recognition of isolated signs. Our architecture uses deep convolutional encoder-decoder architecture …

Sensor-Based Human Activity Recognition Using Bidirectional Lstm For Closely Related Activities, Arumugam Thendramil Pavai

Electronic Theses, Projects, and Dissertations

Recognizing human activities using deep learning methods has significance in many fields such as sports, motion tracking, surveillance, healthcare and robotics. Inertial sensors comprising of accelerometers and gyroscopes are commonly used for sensor based HAR. In this study, a Bidirectional Long Short-Term Memory (BLSTM) approach is explored for human activity recognition and classification for closely related activities on a body worn inertial sensor data that is provided by the UTD-MHAD dataset. The BLSTM model of this study could achieve an overall accuracy of 98.05% for 15 different activities and 90.87% for 27 different activities performed by 8 persons with 4 …

Efficacy Of Deep Learning In Support Of Smart Services, Basheer Mohammed Basheer Qolomany

Dissertations

The massive amount of streaming data generated and captured by smart service appliances, sensors and devices needs to be analyzed by algorithms, transformed into information, and minted to extract knowledge to facilitate timely actions and better decision making. This can lead to new products and services that can dramatically transform our lives. Machine learning and data analytics will undoubtedly play a critical role in enabling the delivery of smart services. Within the machine-learning domain, Deep Learning (DL) is emerging as a superior new approach that is much more effective than any rule or formula used by traditional machine learning. Furthermore, …

Embedded Cyclegan For Shape-Agnostic Image-To-Image Translation, Ram Longman

Theses

Image-to-Image translation is the task of translating images between domains while maintaining the identity of the images. The task can be used for entertainment purposes and applications, data augmentation, semantic image segmentation, and more. Generative Adversarial Networks (GANs), and in particular Conditional GANs have recently shown incredible success in image-to-image translation and semantic manipulation. However, such methods require paired data, meaning that an image must have ground-truth translations across domains. Cycle-consistent GANs solve this problem by using unpaired data. Such methods work well for translations that involve color and texture changes but fail when shape changes are required. This research …

Transfer Learning With Deep Recurrent Neural Networks For Remaining Useful Life Estimation, Ansi Zhang, Shaobo Li, Yuxin Cui, Zhonghao Liu, Guanci Yang, Jianjun Hu

Faculty Publications

Prognostics, such as remaining useful life (RUL) prediction, is a crucial task in condition-based maintenance. A major challenge in data-driven prognostics is the difficulty of obtaining a sufficient number of samples of failure progression. However, for traditional machine learning methods and deep neural networks, enough training data is a prerequisite to train good prediction models. In this work, we proposed a transfer learning algorithm based on Bi-directional Long Short-Term Memory (BLSTM) recurrent neural networks for RUL estimation, in which the models can be first trained on different but related datasets and then fine-tuned by the target dataset. Extensive experimental results …

Transfer Learning With Deep Recurrent Neural Networks For Remaining Useful Life Estimation, Ansi Zhang, Honglei Wang, Shaobo Li, Yuxin Cui, Guanci Yang, Jianjun Hu

Faculty Publications

Prognostics, such as remaining useful life (RUL) prediction, is a crucial task in condition-based maintenance. A major challenge in data-driven prognostics is the difficulty of obtaining a sufficient number of samples of failure progression. However, for traditional machine learning methods and deep neural networks, enough training data is a prerequisite to train good prediction models. In this work, we proposed a transfer learning algorithm based on Bi-directional Long Short-Term Memory (BLSTM) recurrent neural networks for RUL estimation, in which the models can be first trained on different but related datasets and then fine-tuned by the target dataset. Extensive experimental results …

Sdnet2018: An Annotated Image Dataset For Non-Contact Concrete Crack Detection Using Deep Convolutional Neural Networks, Sattar Dorafshan, Robert J. Thomas, Marc Maguire

Civil and Environmental Engineering Faculty Publications

SDNET2018 is an annotated image dataset for training, validation, and benchmarking of artificial intelligence based crack detection algorithms for concrete. SDNET2018 contains over 56,000 images of cracked and non-cracked concrete bridge decks, walls, and pavements. The dataset includes cracks as narrow as 0.06 mm and as wide as 25 mm. The dataset also includes images with a variety of obstructions, including shadows, surface roughness, scaling, edges, holes, and background debris. SDNET2018 will be useful for the continued development of concrete crack detection algorithms based on deep convolutional neural networks (DCNNs), which are a subject of continued research in the field …

Learning Generalized Video Memory For Automatic Video Captioning, Poo-Hee Chang, Ah-Hwee Tan

Research Collection School Of Computing and Information Systems

Recent video captioning methods have made great progress by deep learning approaches with convolutional neural networks (CNN) and recurrent neural networks (RNN). While there are techniques that use memory networks for sentence decoding, few work has leveraged on the memory component to learn and generalize the temporal structure in video. In this paper, we propose a new method, namely Generalized Video Memory (GVM), utilizing a memory model for enhancing video description generation. Based on a class of self-organizing neural networks, GVM’s model is able to learn new video features incrementally. The learned generalized memory is further exploited to decode the …

Improving Knowledge Tracing Model By Integrating Problem Difficulty, Sein Minn, Feida Zhu, Michel C. Desmarais

Research Collection School Of Computing and Information Systems

Intelligent Tutoring Systems (ITS) are designed for providing personalized instructions to students with the needs of their skills. Assessment of student knowledge acquisition dynamically is nontrivial during her learning process with ITS. Knowledge tracing, a popular student modeling technique for student knowledge assessment in adaptive tutoring, which is used for tracing student's knowledge state and detecting student's knowledge acquisition by using decomposed individual skill or problems with a single skill per problem. Unfortunately, recent KT models fail to deal with practices of complex skill composition and variety of concepts included in a problem simultaneously. Our goal is to investigate a …

Multi-Modal Learning Using Deep Neural Networks, Dheeraj Kumar Peri

Theses

Humans have an incredible ability to process and understand information from multiple sources such as images, video, text, and speech. Recent success of deep neural networks has enabled us to develop algorithms which give machines the ability to understand and interpret this information. Convolutional Neural Networks (CNN) have become a standard in extracting rich features from visual stimuli. Recurrent Neural Networks (RNNs) and its variants such as Long Short Term Memory (LSTMs) units have been highly successful in encoding and decoding sequential information like speech and text. Although these networks are highly successful when applied to narrow applications, there is …

Datanet: Deep Learning Based Encrypted Network Traffic Classification In Sdn Home Gateway, Pan Wang, Feng Ye, Xuejiao Chen, Yi Qian

Department of Electrical and Computer Engineering: Faculty Publications

A smart home network will support various smart devices and applications, e.g., home automation devices, E-health devices, regular computing devices, and so on. Most devices in a smart home access the Internet through a home gateway (HGW). In this paper, we propose a software-defined network (SDN)-HGW framework to better manage distributed smart home networks and support the SDN controller of the core network. The SDN controller enables efficient network quality-of-service management based on real-time traffic monitoring and resource allocation of the core network. However, it cannot provide network management in distributed smart homes. Our proposed SDN-HGW extends the control to …

A Critical Thinking Guide For Discussion Board Forums, Jessica Nordness

Culminating Projects in Information Media

This portfolio aims to demonstrate the development of a training guide for discussion board forums for undergraduate students at St. Cloud State University. The purpose of this training guide is to help students understand different types of questioning, as well as, how to critically think and respond to discussion board questions by objectively considering different points of view. To develop background knowledge, the researcher explored topics such as: surface learning versus deeper learning, critical thinking in online learning, higher order thinking and learning, as well as, the role of questions. This portfolio comprises of a training guide that has two …

Deep D-Bar: Real-Time Electrical Impedance Tomography Imaging With Deep Neural Networks, Sarah J. Hamilton, Andreas Hauptmann

Mathematical and Statistical Science Faculty Research and Publications

The mathematical problem for electrical impedance tomography (EIT) is a highly nonlinear ill-posed inverse problem requiring carefully designed reconstruction procedures to ensure reliable image generation. D-bar methods are based on a rigorous mathematical analysis and provide robust direct reconstructions by using a low-pass filtering of the associated nonlinear Fourier data. Similarly to low-pass filtering of linear Fourier data, only using low frequencies in the image recovery process results in blurred images lacking sharp features, such as clear organ boundaries. Convolutional neural networks provide a powerful framework for post-processing such convolved direct reconstructions. In this paper, we demonstrate that these CNN …

Meta Learning For Graph Neural Networks, Rohan N. Dhamdhere

Theses

Deep learning has enabled incredible advances in pattern recognition such as the fields of computer vision and natural language processing. One of the most successful areas of deep learning is Convolutional Neural Networks (CNNs). CNNs have helped improve performance on many difficult video and image understanding tasks but are restricted to dense gridded structures. Further, designing their architectures can be challenging, even for image classification problems. The recently introduced graph CNNs can work on both dense gridded structures as well as generic graphs. Graph CNNs have been performing at par with traditional CNNs on tasks such as point cloud classification …

End-To-End Convolutional Neural Network Model For Gear Fault Diagnosis Based On Sound Signals, Yong Yao, Honglei Wang, Shaobo Li, Zhongnhao Liu, Gui Gui, Yabo Dan, Jianjun Hu

Faculty Publications

Currently gear fault diagnosis is mainly based on vibration signals with a few studies on acoustic signal analysis. However, vibration signal acquisition is limited by its contact measuring while traditional acoustic-based gear fault diagnosis relies heavily on prior knowledge of signal processing techniques and diagnostic expertise. In this paper, a novel deep learning-based gear fault diagnosis method is proposed based on sound signal analysis. By establishing an end-to-end convolutional neural network (CNN), the time and frequency domain signals can be fed into the model as raw signals without feature engineering. Moreover, multi-channel information from different microphones can also be fused …

Toward Audio Beehive Monitoring: Deep Learning Vs. Standard Machine Learning In Classifying Beehive Audio Samples, Vladmir Kulyukin, Sarbajit Mukherjee, Prakhar Amlathe

Computer Science Faculty and Staff Publications

Electronic beehive monitoring extracts critical information on colony behavior and phenology without invasive beehive inspections and transportation costs. As an integral component of electronic beehive monitoring, audio beehive monitoring has the potential to automate the identification of various stressors for honeybee colonies from beehive audio samples. In this investigation, we designed several convolutional neural networks and compared their performance with four standard machine learning methods (logistic regression, k-nearest neighbors, support vector machines, and random forests) in classifying audio samples from microphones deployed above landing pads of Langstroth beehives. On a dataset of 10,260 audio samples where the training and testing …