Physical Sciences and Mathematics Commons^™

Differential Privacy Protection Over Deep Learning: An Investigation Of Its Impacted Factors, Ying Lin, Ling-Yan Bao, Ze-Minghui Li, Shu-Sheng Si, Chao-Hsien Chu

Research Collection School Of Computing and Information Systems

Deep learning (DL) has been widely applied to achieve promising results in many fields, but it still exists various privacy concerns and issues. Applying differential privacy (DP) to DL models is an effective way to ensure privacy-preserving training and classification. In this paper, we revisit the DP stochastic gradient descent (DP-SGD) method, which has been used by several algorithms and systems and achieved good privacy protection. However, several factors, such as the sequence of adding noise, the models used etc., may impact its performance with various degrees. We empirically show that adding noise first and clipping second will not only …

Secure And Verifiable Inference In Deep Neural Networks, Guowen Xu, Hongwei Li, Hao Ren, Jianfei Sun, Shengmin Xu, Jianting Ning, Haoming Yang, Kan Yang, Robert H. Deng

Research Collection School Of Computing and Information Systems

Outsourced inference service has enormously promoted the popularity of deep learning, and helped users to customize a range of personalized applications. However, it also entails a variety of security and privacy issues brought by untrusted service providers. Particularly, a malicious adversary may violate user privacy during the inference process, or worse, return incorrect results to the client through compromising the integrity of the outsourced model. To address these problems, we propose SecureDL to protect the model’s integrity and user’s privacy in Deep Neural Networks (DNNs) inference process. In SecureDL, we first transform complicated non-linear activation functions of DNNs to low-degree …

New Methods For Deep Learning Based Real-Valued Inter-Residue Distance Prediction, Jacob Barger

Theses

Background: Much of the recent success in protein structure prediction has been a result of accurate protein contact prediction--a binary classification problem. Dozens of methods, built from various types of machine learning and deep learning algorithms, have been published over the last two decades for predicting contacts. Recently, many groups, including Google DeepMind, have demonstrated that reformulating the problem as a multi-class classification problem is a more promising direction to pursue. As an alternative approach, we recently proposed real-valued distance predictions, formulating the problem as a regression problem. The nuances of protein 3D structures make this formulation appropriate, allowing predictions …

Deepfrag-K: A Fragment-Based Deep Learning Approach For Protein Fold Recognition, Wessam Elhefnawy, Min Li, Jianxin Wang, Yaohang Li

Computer Science Faculty Publications

Background: One of the most essential problems in structural bioinformatics is protein fold recognition. In this paper, we design a novel deep learning architecture, so-called DeepFrag-k, which identifies fold discriminative features at fragment level to improve the accuracy of protein fold recognition. DeepFrag-k is composed of two stages: the first stage employs a multi-modal Deep Belief Network (DBN) to predict the potential structural fragments given a sequence, represented as a fragment vector, and then the second stage uses a deep convolutional neural network (CNN) to classify the fragment vector into the corresponding fold.

Results: Our results show that DeepFrag-k yields …

Deepcommenter: A Deep Code Comment Generation Tool With Hybrid Lexical And Syntactical Information, Boao Li, Meng Yan, Xin Xia, Xing Hu, Ge Li, David Lo

Research Collection School Of Computing and Information Systems

As the scale of software projects increases, the code comments are more and more important for program comprehension. Unfortunately, many code comments are missing, mismatched or outdated due to tight development schedule or other reasons. Automatic code comment generation is of great help for developers to comprehend source code and reduce their workload. Thus, we propose a code comment generation tool (DeepCommenter) to generate descriptive comments for Java methods. DeepCommenter formulates the comment generation task as a machine translation problem and exploits a deep neural network that combines the lexical and structural information of Java methods. We implement DeepCommenter in …

Experimental Comparison Of Features And Classifiers For Android Malware Detection, Lwin Khin Shar, Biniam Fisseha Demissie, Mariano Ceccato, Wei Minn

Research Collection School Of Computing and Information Systems

Android platform has dominated the smart phone market for years now and, consequently, gained a lot of attention from attackers. Malicious apps (malware) pose a serious threat to the security and privacy of Android smart phone users. Available approaches to detect mobile malware based on machine learning rely on features extracted with static analysis or dynamic analysis techniques. Dif- ferent types of machine learning classi ers (such as support vector machine and random forest) deep learning classi ers (based on deep neural networks) are then trained on extracted features, to produce models that can be used to detect mobile malware. …

Multi-Modal Cooking Workflow Construction For Food Recipes, Liangming Pan, Jingjing Chen, Jianlong Wu, Shaoteng Liu, Chong-Wah Ngo, Min-Yen Kan, Yugang Jiang, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Understanding food recipe requires anticipating the implicit causal effects of cooking actions, such that the recipe can be converted into a graph describing the temporal workflow of the recipe. This is a non-trivial task that involves common-sense reasoning. However, existing efforts rely on hand-crafted features to extract the workflow graph from recipes due to the lack of large-scale labeled datasets. Moreover, they fail to utilize the cooking images, which constitute an important part of food recipes. In this paper, we build MM-ReS, the first large-scale dataset for cooking workflow construction, consisting of 9,850 recipes with human-labeled workflow graphs. Cooking steps …

Unveiling The Molecular Mechanism Of Sars-Cov-2 Main Protease Inhibition From 137 Crystal Structures Using Algebraic Topology And Deep Learning, Duc Duy Nguyen, Kaifu Gao, Jiahui Chen, Rui Wang, Guo-Wei Wei

Mathematics Faculty Publications

Currently, there is neither effective antiviral drugs nor vaccine for coronavirus disease 2019 (COVID-19) caused by acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Due to its high conservativeness and low similarity with human genes, SARS-CoV-2 main protease (M^pro) is one of the most favorable drug targets. However, the current understanding of the molecular mechanism of M^pro inhibition is limited by the lack of reliable binding affinity ranking and prediction of existing structures of M^pro-inhibitor complexes. This work integrates mathematics (i.e., algebraic topology) and deep learning (MathDL) to provide a reliable ranking of the binding …

Semantic-Driven Unsupervised Image-To-Image Translation For Distinct Image Domains, Wesley Ackerman

Theses and Dissertations

We expand the scope of image-to-image translation to include more distinct image domains, where the image sets have analogous structures, but may not share object types between them. Semantic-Driven Unsupervised Image-to-Image Translation for Distinct Image Domains (SUNIT) is built to more successfully translate images in this setting, where content from one domain is not found in the other. Our method trains an image translation model by learning encodings for semantic segmentations of images. These segmentations are translated between image domains to learn meaningful mappings between the structures in the two domains. The translated segmentations are then used as the basis …

Joint 1d And 2d Neural Networks For Automatic Modulation Recognition, Luis M. Rosario Morel

Theses and Dissertations

The digital communication and radar community has recently manifested more interest in using data-driven approaches for tasks such as modulation recognition, channel estimation and distortion correction. In this research we seek to apply an object detector for parameter estimation to perform waveform separation in the time and frequency domain prior to classification. This enables the full automation of detecting and classifying simultaneously occurring waveforms. We leverage a lD ResNet implemented by O'Shea et al. in [1] and the YOLO v3 object detector designed by Redmon et al. in [2]. We conducted an in depth study of the performance of these …

Physics-Constrained Hyperspectral Data Exploitation Across Diverse Atmospheric Scenarios, Nicholas M. Westing

Theses and Dissertations

Hyperspectral target detection promises new operational advantages, with increasing instrument spectral resolution and robust material discrimination. Resolving surface materials requires a fast and accurate accounting of atmospheric effects to increase detection accuracy while minimizing false alarms. This dissertation investigates deep learning methods constrained by the processes governing radiative transfer to efficiently perform atmospheric compensation on data collected by long-wave infrared (LWIR) hyperspectral sensors. These compensation methods depend on generative modeling techniques and permutation invariant neural network architectures to predict LWIR spectral radiometric quantities. The compensation algorithms developed in this work were examined from the perspective of target detection performance using …

A Semi-Automated Approach To Medical Image Segmentation Using Conditional Random Field Inference, Yu-Chi Hu

Dissertations, Theses, and Capstone Projects

Medical image segmentation plays a crucial role in delivering effective patient care in various diagnostic and treatment modalities. Manual delineation of target volumes and all critical structures is a very tedious and highly time-consuming process and introduce uncertainties of treatment outcomes of patients. Fully automatic methods holds great promise for reducing cost and time, while at the same time improving accuracy and eliminating expert variability, yet there are still great challenges. Legally and ethically, human oversight must be integrated with ”smart tools” favoring a semi-automatic technique which can leverage the best aspects of both human and computer.

In this work …

The Gap Of Semantic Parsing: A Survey On Automatic Math Word Problem Solvers, Dongxiang Zhang, Lei Wang, Luming Zhang, Bing Tian Dai, Heng Tao Shen

Research Collection School Of Computing and Information Systems

Solving mathematical word problems (MWPs) automatically is challenging, primarily due to the semantic gap between human-readable words and machine-understandable logics. Despite the long history dated back to the 1960s, MWPs have regained intensive attention in the past few years with the advancement of Artificial Intelligence (AI). Solving MWPs successfully is considered as a milestone towards general AI. Many systems have claimed promising results in self-crafted and small-scale datasets. However, when applied on large and diverse datasets, none of the proposed methods in the literature achieves high precision, revealing that current MWP solvers still have much room for improvement. This motivated …

Analyzing And Assisting Patient Decision-Making In Online Health Communities, Mingda Li

Dissertations

In recent years, many users have joined online health communities (OHC) to seek information, suggestions and social support. However, little has been studied about the roles of OHCs in patients' decision-making processes. The aim of this research is to analyze OHC data to better understand patient decision-making processes, and to provide assistance to OHC users in their decision-making.

In order to analyze or assist patients in their decision-making, a novel classification model is designed to identify discussion threads in OHC that are related to decision making. This is achieved by building a two-step combined deep learning model. Empirical evaluation shows …

Accurate Patient Health Condition Identification And Risk Prediction From Health Data, Jinhe Shi

Dissertations

Online Health communities (OHCs), Electronic Health Records (EHRs) and Claims data contain rich patient information. Accurate identification of patient's health conditions is critical for providing effective and safe health care service and for trustable knowledge discovery. This dissertation proposes several deep-learning based techniques on identifying patient health conditions and predicting adverse drug event (ADE) risks using various health data.

First, this dissertation presents a novel patient experience mining model to differentiate the data expressing user experiences from the one describing hearsays. Indeed, OHCs contain information with varying degrees of quality. Identifying information that describes patient health experience from OHCs is …

Spoken Language Recognition On Open-Source Datasets, Brady Arendale, Samira Zarandioon, Ryan Goodwin, Douglas Reynolds

SMU Data Science Review

The field of speaker and language recognition is constantly being researched and developed, but much of this research is done on private or expensive datasets, making the field more inaccessible than many other areas of machine learning. In addition, many papers make performance claims without comparing their models to other recent research. With the recent development of public multilingual speech corpora such as Mozilla's Common Voice as well as several single-language corpora, we now have the resources to attempt to address both of these problems. We construct an eight-language dataset from Common Voice and a Google Bengali corpus as well …

Machine Learning And Deep Learning Based Entity Resolution Approaches For Unstructured References, Xinming Li

Theses and Dissertations

As a fundamental task in data integration and data quality, Entity Resolution (ER) has been investigated for decades in various domains. The emerging volume of heterogeneously structured data, and even unstructured data, poses a challenge to traditional ER methods. This research is to explore machine learning and deep learning approach to address the challenge from unstructured references data. This research starts with pairwise matching, the core function of all ER tasks. Based on the similarity score vector derived from our designed similarity measurement tool, scoring matrix, machine leaning enhances the performance significantly compared to the manually threshold method. Without similarity …

Deep Learning To Predict Ocean Seabed Type And Source Parameters, David Franklin Van Komen

Theses and Dissertations

In the ocean, light from the surface dissipates quickly leaving sound the only way to see at a distance. Different sediment types on the ocean floor and water properties like salinity, temperature, and ocean depth all change how sound travels across long distances. Hard sediment types, such as sand and bedrock, are highly reflective while softer sediment types, such as mud, are more absorptive and change the received sound upon arrival. Unfortunately, the vast majority of the ocean floor is not mapped and the expenses involved in creating such a map are far too great. Traditional signal processing methods in …

Exploring The Efficacy Of Transfer Learning In Mining Image‑Based Software Artifacts, Natalie Best, Jordan Ott, Erik J. Linstead

Engineering Faculty Articles and Research

Background

Transfer learning allows us to train deep architectures requiring a large number of learned parameters, even if the amount of available data is limited, by leveraging existing models previously trained for another task. In previous attempts to classify image-based software artifacts in the absence of big data, it was noted that standard off-the-shelf deep architectures such as VGG could not be utilized due to their large parameter space and therefore had to be replaced by customized architectures with fewer layers. This proves to be challenging to empirical software engineers who would like to make use of existing architectures without …

Empirical Studies Of Deep Learning On Information Diffusion On Social Networks And Collective Task Learning For Swarm Robotics, Trung T. Nguyen

Dissertations

Researchers in multiple disciplines have recently adopted deep learning because of its ability of high accuracy representation learning from big and complex data. My research goal in this thesis is developing deep learning models for information diffusion analysis on social networks and collective tasks learning in swarm robotics. Firstly, the information diffusion on social networks is modeled as a multivariate time series in three dimensions with ten features. Then, we applied time-series clustering algorithms with Dynamic Time Warping to discover different patterns of our models. Then, we build a prediction model based on LSTM, which outperforms traditional time-series prediction methods. …

Applications Of Artificial Intelligence And Graphy Theory To Cyberbullying, Jesse D. Simpson

Graduate Theses/Dissertations

Cyberbullying is an ongoing and devastating issue in today's online social media. Abusive users engage in cyber-harassment by utilizing social media to send posts, private messages, tweets, or pictures to innocent social media users. Detecting and preventing cases of cyberbullying is crucial. In this work, I analyze multiple machine learning, deep learning, and graph analysis algorithms and explore their applicability and performance in pursuit of a robust system for detecting cyberbullying. First, I evaluate the performance of the machine learning algorithms Support Vector Machine, Naïve Bayes, Random Forest, Decision Tree, and Logistic Regression. This yielded positive results and obtained upwards …

Novel Deep Learning Methods Combined With Static Analysis For Source Code Processing, Duy Quoc Nghi Bui

Dissertations and Theses Collection (Open Access)

It is desirable to combine machine learning and program analysis so that one can leverage the best of both to increase the performance of software analytics. On one side, machine learning can analyze the source code of thousands of well-written software projects that can uncover patterns that partially characterize software that is reliable, easy to read, and easy to maintain. On the other side, the program analysis can be used to define rigorous and unique rules that are only available in programming languages, which enrich the representation of source code and help the machine learning to capture the patterns better. …

Deep Learning For Remote Sensing Image Processing, Yan Lu

Computational Modeling & Simulation Engineering Theses & Dissertations

Remote sensing images have many applications such as ground object detection, environmental change monitoring, urban growth monitoring and natural disaster damage assessment. As of 2019, there were roughly 700 satellites listing “earth observation” as their primary application. Both spatial and temporal resolutions of satellite images have improved consistently in recent years and provided opportunities in resolving fine details on the Earth's surface. In the past decade, deep learning techniques have revolutionized many applications in the field of computer vision but have not fully been explored in remote sensing image processing. In this dissertation, several state-of-the-art deep learning models have been …

Secure Mobile Computing By Using Convolutional And Capsule Deep Neural Networks, Rui Ning

Electrical & Computer Engineering Theses & Dissertations

Mobile devices are becoming smarter to satisfy modern user's increasing needs better, which is achieved by equipping divers of sensors and integrating the most cutting-edge Deep Learning (DL) techniques. As a sophisticated system, it is often vulnerable to multiple attacks (side-channel attacks, neural backdoor, etc.). This dissertation proposes solutions to maintain the cyber-hygiene of the DL-Based smartphone system by exploring possible vulnerabilities and developing countermeasures.

First, I actively explore possible vulnerabilities on the DL-Based smartphone system to develop proactive defense mechanisms. I discover a new side-channel attack on smartphones using the unrestricted magnetic sensor data. I demonstrate that attackers can …

Facial Expression Recognition In The Wild Using Convolutional Neural Networks, Amir Hossein Farzaneh

All Graduate Theses and Dissertations, Spring 1920 to Summer 2023

Facial Expression Recognition (FER) is the task of predicting a specific facial expression given a facial image. FER has demonstrated remarkable progress due to the advancement of deep learning. Generally, a FER system as a prediction model is built using two sub-modules: 1. Facial image representation model that learns a mapping from the input 2D facial image to a compact feature representation in the embedding space, and 2. A classifier module that maps the learned features to the label space comprising seven labels of neutral, happy, sad, surprise, anger, fear, or disgust. Ultimately, …

Facing The Hard Problems In Fgvc, Connor Stanley Anderson

Theses and Dissertations

In fine-grained visual categorization (FGVC), there is a near-singular focus in pursuit of attaining state-of-the-art (SOTA) accuracy. This work carefully analyzes the performance of recent SOTA methods, quantitatively, but more importantly, qualitatively. We show that these models universally struggle with certain "hard" images, while also making complementary mistakes. We underscore the importance of such analysis, and demonstrate that combining complementary models can improve accuracy on the popular CUB-200 dataset by over 5%. In addition to detailed analysis and characterization of the errors made by these SOTA methods, we provide a clear set of recommended directions for future FGVC researchers.

A Survey On Visual Slam Based On Deep Learning, Ruijun Liu, Xiangshang Wang, Zhang Chen, Bohua Zhang

Journal of System Simulation

Abstract: Following the development of computer vision and robotics, visual Simultaneous Localization and Mapping becomes a research focus in the field of unmanned systems. The powerful advantages of deep learning in the image processing offer a huge opportunity to the wide combination of the two fields. The outstanding research achievements of deep learning combined with visual odometry, loop closure detection and semantic Simultaneous Localization and Mapping are summarized. A comparison between the traditional algorithm and method based on deep learning is carried out. The development direction of visual Simultaneous Localization and Mapping based on deep learning is …

Human Depth Maps Restoration Based On Guided Gan, Jingfang Yin, Dengming Zhu, Shi Min, Zhaoqi Wang

Journal of System Simulation

Abstract: The depth maps captured by a small depth camera on mobile devices suffer from the problem of severe holes. The Guided Generative Adversarial Network (Guided GAN) based on deep learning is proposed to restore human depth maps with above problems. The high-precision human segmentation features and depth class features are extracted from the monocular RGB image by the guider based on the stacked hourglass network. The holes in the human depth maps are filled by the special generator under the guidance of the extracted human features. In order to get the more realistic results, the discriminator is introduced …

Deep Learning For Identifying Breast Cancer, Yihong Li

Master of Science in Computer Science Theses

Medical images are playing an increasingly important role in the prevention and diagnosis of diseases. Medical images often contain massive amounts of data. Professional interpretation usually requires a long time of professional study and experience accumulation by doctors. Therefore, the use of super storage and computing power in deep learning as a basis can effectively process a large amount of medical data. Breast cancer brings great harm to female patients, and early diagnosis is the most effective prevention and treatment method, so this project will create a new optimized breast cancer auxiliary diagnosis model based on ResNet. Analyze and process, …

Deep Learning For Identifying Lung Diseases, Lin Wang

Master of Science in Computer Science Theses

Growing health problems, such as lung diseases, especially for children and the elderly, require better diagnostic methods, such as computer-based solutions, and it is crucial to detect and treat these problems early. The purpose of this article is to design and implement a new computer vision-based algorithm based on lung disease diagnosis, which has better performance in lung disease recognition than previous models to reduce lung-related health problems and costs . In addition, we have improved the accuracy of the five lung diseases detection, which helps doctors and doctors use computers to solve this problem at an early stage.