Engineering Commons^™

Improving Credit Card Fraud Detection Using Transfer Learning And Data Resampling Techniques, Charmaine Eunice Mena Vinarta

Electronic Theses, Projects, and Dissertations

This Culminating Experience Project explores the use of machine learning algorithms to detect credit card fraud. The research questions are: Q1. What cross-domain techniques developed in other domains can be effectively adapted and applied to mitigate or eliminate credit card fraud, and how do these techniques compare in terms of fraud detection accuracy and efficiency? Q2. To what extent do synthetic data generation methods effectively mitigate the challenges posed by imbalanced datasets in credit card fraud detection, and how do these methods impact classification performance? Q3. To what extent can the combination of transfer learning and innovative data resampling techniques …

Toward Generating Efficient Deep Neural Networks, Chengcheng Li

Doctoral Dissertations

Recent advances in deep neural networks have led to tremendous applications in various tasks, such as object classification and detection, image synthesis, natural language processing, game playing, and biological imaging. However, deploying these pre-trained networks on resource-limited devices poses a challenge, as most state-of- the-art networks contain millions of parameters, making them cumbersome and slow in real-world applications. To address this problem, numerous network compression and acceleration approaches, also known as efficient deep neural networks or efficient deep learning, have been investigated, in terms of hardware and software (algorithms), training, and inference. The aim of this dissertation is to study …

Improving Classification In Single And Multi-View Images, Hadi Kanaan Hadi Salman

Graduate Theses and Dissertations

Image classification is a sub-field of computer vision that focuses on identifying objects within digital images. In order to improve image classification we must address the following areas of improvement: 1) Single and Multi-View data quality using data pre-processing techniques. 2) Enhancing deep feature learning to extract alternative representation of the data. 3) Improving decision or prediction of labels. This dissertation presents a series of four published papers that explore different improvements of image classification. In our first paper, we explore the Siamese network architecture to create a Convolution Neural Network based similarity metric. We learn the priority features that …

Self-Learning Algorithms For Intrusion Detection And Prevention Systems (Idps), Juan E. Nunez, Roger W. Tchegui Donfack, Rohit Rohit, Hayley Horn

SMU Data Science Review

Today, there is an increased risk to data privacy and information security due to cyberattacks that compromise data reliability and accessibility. New machine learning models are needed to detect and prevent these cyberattacks. One application of these models is cybersecurity threat detection and prevention systems that can create a baseline of a network's traffic patterns to detect anomalies without needing pre-labeled data; thus, enabling the identification of abnormal network events as threats. This research explored algorithms that can help automate anomaly detection on an enterprise network using Canadian Institute for Cybersecurity data. This study demonstrates that Neural Networks with Bayesian …

Application Of Shallow Neural Networks To Retail Intermittent Demand Time Series, Urko Allende

Dissertations

Accurate sales predictions are essential for businesses in the fast-moving consumer goods (FMCG) industry. However, their demand forecasts are often unreliable, leading to imprecisions that affect downstream decisions. This dissertation proposes using an artificial neural network to improve intermittent demand forecasting in the retail sector. The research investigates the validity of using unprocessed historical information, eluding hand-crafted features, to learn patterns in intermittent demand data. The experiment tests a selection of shallow neural network architectures that can expedite the time-to-market in comparison to conventional demand forecasting methods. The results demonstrate that organisations that still rely on manual and direct forecasting …

Analyzing Ground Motion Records With Cvi Fuzzy Art, Dustin Tanksley, Xinzhe Yuan, Genda Chen, Donald C. Wunsch

Civil, Architectural and Environmental Engineering Faculty Research & Creative Works

This paper explores using Cluster Validity Indices Fuzzy Adaptative Resonance Theory (CVI Fuzzy ART) to cluster ground motion records (GMRs). Clustering the features extracted from a supervised network trained for predicting the structure damage results in less overfitting from the trained network. Using Cluster Validity Indices (CVIs) to evaluate the clustering gives feedback to how well the data is being classified, allowing further separation of the data. By using CVI Fuzzy ART in combination with features extracted from a trained Convolutional Neural Network (CNN), we were able to form additional clusters in the data. Within the primary clusters, accuracy was …

Recurrent Neural Networks For Flash Gdp Estimates In Ireland: A Comparison With Traditional Econometric Methods, Justin Flannery

ICT

GDP is the single most important barometer for the health of an economy. It’s an important input into the decision making processes of government, industry and state institutions such as central banks. To be useful as an indicator, GDP estimates need to be both timely and accurate. To meet the needs of users, many national statistical institutes publish early or flash estimates of GDP which are produced within 30 days after the end of a quarter. Given the long lags involved in the data collection processes which feed into GDP estimates, these flash estimates are often largely model based. Within …

Quadratic Neural Network Architecture As Evaluated Relative To Conventional Neural Network Architecture, Reid Taylor

Senior Theses

Current work in the field of deep learning and neural networks revolves around several variations of the same mathematical model for associative learning. These variations, while significant and exceptionally applicable in the real world, fail to push the limits of modern computational prowess. This research does just that: by leveraging high order tensors in place of 2nd order tensors, quadratic neural networks can be developed and can allow for substantially more complex machine learning models which allow for self-interactions of collected and analyzed data. This research shows the theorization and development of mathematical model necessary for such an idea to …

Autonomous Rock Segmentation From Lidar Point Clouds Using Machine Learning Approaches, Lauren E. Flanagan

Electronic Thesis and Dissertation Repository

Rover navigation on planetary surfaces currently uses a method called blind drive which requires a navigation goal as input from operators on Earth and uses camera images to autonomously detect obstacles. Images can be affected by lighting conditions, are not highly accurate from far distances, and will not work in the dark; these factors negatively impact the autonomous capabilities of rovers. By improving a rover's ability to autonomously detect obstacles, the capabilities of rovers in future missions would improve; for example, enabling exploration of permanently shadowed regions, and allowing faster driving speeds and farther travel distances. This thesis demonstrates how …

Deepfakes Generated By Generative Adversarial Networks, Olympia A. Paul

Honors College Theses

Deep learning is a type of Artificial Intelligence (AI) that mimics the workings of the human brain in processing data such as speech recognition, visual object recognition, object detection, language translation, and making decisions. A Generative adversarial network (GAN) is a special type of deep learning, designed by Goodfellow et al. (2014), which is what we call convolution neural networks (CNN). How a GAN works is that when given a training set, they can generate new data with the same information as the training set, and this is often what we refer to as deep fakes. CNN takes an input …

Adaptive Mobile Eeg Noise Cancellation Using 2d Convolutional Autoencoders For Bci Authentication, Tyree Lewis

USF Tampa Graduate Theses and Dissertations

Electroencephalography (EEG) signals can be used for many purposes and has the potential to be adapted to various systems. When EEG is recorded from users, these studies are performed primarily in an indoor environment, while the user is stationary. This is due to the levels of noise that are experienced when recording EEG data, to minimize errors in the data. This thesis aims to adapt tasks that are performed indoors to an external environment by removing both noise and artefacts in EEG, using a 2D Convolutional Autoencoder (CAE). The data is recorded from subjects is passed into the 2D CAE …

Low-Power And Reconfigurable Asynchronous Asic Design Implementing Recurrent Neural Networks, Spencer Nelson

Graduate Theses and Dissertations

Artificial intelligence (AI) has experienced a tremendous surge in recent years, resulting in high demand for a wide array of implementations of algorithms in the field. With the rise of Internet-of-Things devices, the need for artificial intelligence algorithms implemented in hardware with tight design restrictions has become even more prevalent. In terms of low power and area, ASIC implementations have the best case. However, these implementations suffer from high non-recurring engineering costs, long time-to-market, and a complete lack of flexibility, which significantly hurts their appeal in an environment where time-to-market is so critical. The time-to-market gap can be shortened through …

Identifying Significant Features For Player Evaluation In Nfl Comparing Anns And Traditional Models, Ronan Walsh

Dissertations

The evaluation of player performance in sports is popular and important in modern sports, enabling teams to use real data in the construction of their rosters. This dissertation proposes to apply machine learning algorithms to predicting the player evaluations from a leading NFL analytics company who use a combination of statistics and expert evaluation. In addition, it will investigate what features are significant in the evaluation of a position. Data for the dissertation is obtained from multiple online sources - Pro Football Reference and Pro Football Focus (the the NFL analytics company). These data sets are combined and analysed before …

Short Term Energy Forecasting For A Microgird Load Using Lstm Rnn, Akhil Soman

Masters Theses

Decentralization of the electric grid can increase resiliency (during natural disasters) and can reduce T&D energy losses and emissions. Microgrids and DERs can enable this to happen. It is important to optimally control microgrids and DERs to extract the greatest economic, environmental and resiliency benefits. This is enabled by robust forecasting to optimally control loads and energy sources. An integral part of microgrid control is power side and load side demand forecasting.

In this thesis, we look at the ability of a powerful neural network algorithm to forecast the load side demand for a microgrid using the UMass campus as …

Scalable Profiling And Visualization For Characterizing Microbiomes, Camilo Valdes

FIU Electronic Theses and Dissertations

Metagenomics is the study of the combined genetic material found in microbiome samples, and it serves as an instrument for studying microbial communities, their biodiversities, and the relationships to their host environments. Creating, interpreting, and understanding microbial community profiles produced from microbiome samples is a challenging task as it requires large computational resources along with innovative techniques to process and analyze datasets that can contain terabytes of information.

The community profiles are critical because they provide information about what microorganisms are present in the sample, and in what proportions. This is particularly important as many human diseases and environmental disasters …

Aspect And Opinion Aware Abstractive Review Summarization With Reinforced Hard Typed Decoder, Yufei Tian, Jianfei Yu, Jing Jiang

Research Collection School Of Computing and Information Systems

In this paper, we study abstractive review summarization. Observing that review summaries often consist of aspect words, opinion words and context words, we propose a two-stage reinforcement learning approach, which first predicts the output word type from the three types, and then leverages the predicted word type to generate the final word distribution. Experimental results on two Amazon product review datasets demonstrate that our method can consistently outperform several strong baseline approaches based on ROUGE scores.

Utilizing Trajectory Optimization In The Training Of Neural Network Controllers, Nicholas Kimball

Master's Theses

Applying reinforcement learning to control systems enables the use of machine learning to develop elegant and efficient control laws. Coupled with the representational power of neural networks, reinforcement learning algorithms can learn complex policies that can be difficult to emulate using traditional control system design approaches. In this thesis, three different model-free reinforcement learning algorithms, including Monte Carlo Control, REINFORCE with baseline, and Guided Policy Search are compared in simulated, continuous action-space environments. The results show that the Guided Policy Search algorithm is able to learn a desired control policy much faster than the other algorithms. In the inverted pendulum …

Labeling Paths With Convolutional Neural Networks, Sean Wallace, Kyle Wuerch

Computer Engineering

With the increasing development of autonomous vehicles, being able to detect driveable paths in arbitrary environments has become a prevalent problem in multiple industries. This project explores a technique which utilizes a discretized output map that is used to color an image based on the confidence that each block is a driveable path. This was done using a generalized convolutional neural network that was trained on a set of 3000 images taken from the perspective of a robot along with matching masks marking which portion of the image was a driveable path. The techniques used allowed for a labeling accuracy …

Neuroengineering Of Clustering Algorithms, Leonardo Enzo Brito Da Silva

Doctoral Dissertations

"Cluster analysis can be broadly divided into multivariate data visualization, clustering algorithms, and cluster validation. This dissertation contributes neural network-based techniques to perform all three unsupervised learning tasks. Particularly, the first paper provides a comprehensive review on adaptive resonance theory (ART) models for engineering applications and provides context for the four subsequent papers. These papers are devoted to enhancements of ART-based clustering algorithms from (a) a practical perspective by exploiting the visual assessment of cluster tendency (VAT) sorting algorithm as a preprocessor for ART offline training, thus mitigating ordering effects; and (b) an engineering perspective by designing a family of …

Deep Neural Network Architectures For Modulation Classification Using Principal Component Analysis, Sharan Ramjee, Shengtai Ju, Diyu Yang, Aly El Gamal

The Summer Undergraduate Research Fellowship (SURF) Symposium

In this work, we investigate the application of Principal Component Analysis to the task of wireless signal modulation recognition using deep neural network architectures. Sampling signals at the Nyquist rate, which is often very high, requires a large amount of energy and space to collect and store the samples. Moreover, the time taken to train neural networks for the task of modulation classification is large due to the large number of samples. These problems can be drastically reduced using Principal Component Analysis, which is a technique that allows us to reduce the dimensionality or number of features of the samples …

Lionfish Detection System, Carmelo Furlan, Andrew Boniface

Computer Engineering

Deep neural networks have proven to be an effective method in classification of images. The ability to recognize objects has opened the door for many new systems which use image classification to solve challenging problems where conventional image classification would be inadequate. We trained a large, deep convolutional neural network to identify lionfish from other species that might be found in the same habitats. Google’s Inception framework served as a powerful platform for our fish recognition system. By using transfer learning, we were able to obtain exceptional results for the classification of different species of fish. The convolutional neural network …

Object Tracking In Games Using Convolutional Neural Networks, Anirudh Venkatesh

Master's Theses

Computer vision research has been growing rapidly over the last decade. Recent advancements in the field have been widely used in staple products across various industries. The automotive and medical industries have even pushed cars and equipment into production that use computer vision. However, there seems to be a lack of computer vision research in the game industry. With the advent of e-sports, competitive and casual gaming have reached new heights with regard to players, viewers, and content creators. This has allowed for avenues of research that did not exist prior.

In this thesis, we explore the practicality of object …

Training Neural Networks To Pilot Autonomous Vehicles: Scaled Self-Driving Car, Jason Zisheng Chang

Senior Projects Spring 2018

This project explores the use of deep convolutional neural networks in autonomous cars. Successful implementation of autonomous vehicles has many societal benefits. One of the main benefits is its potential to significantly reduce traffic accidents. In the United States, the National Highway Traffic Safety Administration states that human error is at fault for 93% of automotive crashes. Robust driverless vehicles can prevent many of these collisions. The main challenge in developing autonomous vehicles today is how to create a system that is able to accurately perceive and process the world around it. In 2016, NVIDIA successfully trained a deep convolutional …

Operating System Identification By Ipv6 Communication Using Machine Learning Ensembles, Adrian Ordorica

Graduate Theses and Dissertations

Operating system (OS) identification tools, sometimes called fingerprinting tools, are essential for the reconnaissance phase of penetration testing. While OS identification is traditionally performed by passive or active tools that use fingerprint databases, very little work has focused on using machine learning techniques. Moreover, significantly more work has focused on IPv4 than IPv6. We introduce a collaborative neural network ensemble that uses a unique voting system and a random forest ensemble to deliver accurate predictions. This approach uses IPv6 features as well as packet metadata features for OS identification. Our experiment shows that our approach is valid and we achieve …

Deep Neural Networks As Time Series Forecasters Of Energy Demand, Gregory Merkel

Master's Theses (2009 -)

Short-term load forecasting is important for the day-to-day operation of natural gas utilities. Traditionally, short-term load forecasting of natural gas is done using linear regression, autoregressive integrated moving average models, and artificial neural networks. Many purchasing and operating decisions are made using these forecasts, and there can be high cost to both natural gas utilities and their customers if the short-term load forecast is inaccurate. Therefore, the GasDay lab continues to explore new ways to make better forecasts. Recently, deep neural networks (DNNs) have emerged as a powerful tool in machine learning problems. DNNs have been shown to greatly outperform …

Critical Comparison Of The Classification Ability Of Deep Convolutional Neural Network Frameworks With Support Vector Machine Techniques In The Image Classification Process, Robert Kelly

Dissertations

Recently, a number of new image classification models have been developed to diversify the number of options available to prospective machine learning classifiers, such as Deep Learning. This is particularly important in the field of medical image classification as a misdiagnosis could have a severe impact on the patient. However, an assessment on the level to which a deep learning based Convolutional Neural Network can outperform a Support Vector Machine has not been discussed. In this project, the use of CNN and SVM classifiers is used on a dataset of approx. 55,000 images. This dataset was used to assess the …

Towards Improving Visqol (Virtual Speech Quality Objective Listener) Using Machine Learning Techniques, Joseph Mcnally

Dissertations

Vast amounts of sound data are transmitted every second over digital networks. VoIP services and cellular networks transmit speech data in increasingly greater volumes. Objective sound quality models provide an essential function to measure the quality of this data in real-time. However, these models can suffer from a lack of accuracy with various degradations over networks. This research uses machine learning techniques to create one support vector regression and three neural network mapping models for use with ViSQOLAudio. Each of the mapping models (including ViSQOL and ViSQOLAudio) are tested against two separate speech datasets in order to comparatively study accuracy …

“My Logic Is Undeniable”: Replicating The Brain For Ideal Artificial Intelligence, Samuel C. Adams

Senior Honors Theses

Alan Turing asked if machines can think, but intelligence is more than logic and reason. I ask if a machine can feel pain or joy, have visions and dreams, or paint a masterpiece. The human brain sets the bar high, and despite our progress, artificial intelligence has a long way to go. Studying neurology from a software engineer’s perspective reveals numerous uncanny similarities between the functionality of the brain and that of a computer. If the brain is a biological computer, then it is the embodiment of artificial intelligence beyond anything we have yet achieved, and its architecture is advanced …

The Role Of Uncertainty In Categorical Perception Utilizing Statistical Learning In Robots, Nathaniel V. Powell

Graduate College Dissertations and Theses

At the heart of statistical learning lies the concept of uncertainty.

Similarly, embodied agents such as robots

and animals must likewise address uncertainty, as sensation

is always only a partial reflection of reality. This

thesis addresses the role that uncertainty can play in

a central building block of intelligence: categorization.

Cognitive agents are able to perform tasks like categorical perception

through physical interaction (active categorical perception; ACP),

or passively at a distance (distal categorical perception; DCP).

It is possible that the former scaffolds the learning of

the latter. However, it is unclear whether DCP indeed scaffolds

ACP in humans and …

Neuron Clustering For Mitigating Catastrophic Forgetting In Supervised And Reinforcement Learning, Benjamin Frederick Goodrich

Doctoral Dissertations

Neural networks have had many great successes in recent years, particularly with the advent of deep learning and many novel training techniques. One issue that has affected neural networks and prevented them from performing well in more realistic online environments is that of catastrophic forgetting. Catastrophic forgetting affects supervised learning systems when input samples are temporally correlated or are non-stationary. However, most real-world problems are non-stationary in nature, resulting in prolonged periods of time separating inputs drawn from different regions of the input space.

Reinforcement learning represents a worst-case scenario when it comes to precipitating catastrophic forgetting in neural networks. …