Matching Items (98)
Filtering by
- All Subjects: Machine Learning
- Creators: Computer Science and Engineering Program
- Creators: Electrical Engineering Program
Description
This paper is centered on the use of generative adversarial networks (GANs) to convert or generate RGB images from grayscale ones. The primary goal is to create sensible and colorful versions of a set of grayscale images by training a discriminator to recognize failed or generated images and training a generator to attempt to satisfy the discriminator. The network design is described in further detail below; however there are several potential issues that arise including the averaging of a color for certain images such that small details in an image are not assigned unique colors leading to a neutral blend. We attempt to mitigate this issue as much as possible.
ContributorsMarkabawi, Jah (Co-author) / Masud, Abdullah (Co-author) / Lobo, Ian (Co-author) / Koleber, Keith (Co-author) / Yang, Yingzhen (Thesis director) / Wang, Yancheng (Committee member) / Computer Science and Engineering Program (Contributor, Contributor) / Barrett, The Honors College (Contributor)
Created2021-05
Description
This paper is centered on the use of generative adversarial networks (GANs) to convert or generate RGB images from grayscale ones. The primary goal is to create sensible and colorful versions of a set of grayscale images by training a discriminator to recognize failed or generated images and training a generator to attempt to satisfy the discriminator. The network design is described in further detail below; however there are several potential issues that arise including the averaging of a color for certain images such that small details in an image are not assigned unique colors leading to a neutral blend. We attempt to mitigate this issue as much as possible.
ContributorsMasud, Abdullah Bin (Co-author) / Koleber, Keith (Co-author) / Lobo, Ian (Co-author) / Markabawi, Jah (Co-author) / Yang, Yingzhen (Thesis director) / Wang, Yancheng (Committee member) / Computer Science and Engineering Program (Contributor, Contributor) / Barrett, The Honors College (Contributor)
Created2021-05
Description
Video games often feature agents that the human player interacts with to overcome.
Designing these agents to cover every case of human interaction is difficult, and usually
imperfect, as human players are capable of learning to overcome these agents in unintended
ways. Artificial intelligence is a growing field that seeks to solve problems by simulating
learning in specific environments. The aim of this paper is to explore the applications that the
self play learning branch of artificial intelligence may pose on game development in the future,
and to attempt to implement a working version of a self play agent learning to play a Pokemon
battle. Originally designed Pokemon battle behavior is often suboptimal, getting stuck making
ineffective or incorrect choices, so training a self play model to learn the strategy and structure of
Pokemon battles from a clean slate would result in an organic agent that would outperform the
original behavior of the computer controlled agents. Though unsuccessful in my implementation,
this paper serves as a record of the exploration of this field, and a log of what worked and what
did not, in order to benefit any future person interested in the same topics.
Designing these agents to cover every case of human interaction is difficult, and usually
imperfect, as human players are capable of learning to overcome these agents in unintended
ways. Artificial intelligence is a growing field that seeks to solve problems by simulating
learning in specific environments. The aim of this paper is to explore the applications that the
self play learning branch of artificial intelligence may pose on game development in the future,
and to attempt to implement a working version of a self play agent learning to play a Pokemon
battle. Originally designed Pokemon battle behavior is often suboptimal, getting stuck making
ineffective or incorrect choices, so training a self play model to learn the strategy and structure of
Pokemon battles from a clean slate would result in an organic agent that would outperform the
original behavior of the computer controlled agents. Though unsuccessful in my implementation,
this paper serves as a record of the exploration of this field, and a log of what worked and what
did not, in order to benefit any future person interested in the same topics.
ContributorsCiudad, Erick Marcel (Author) / Meuth, Ryan (Thesis director) / Kobayashi, Yoshihiro (Committee member) / Computing and Informatics Program (Contributor) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2020-12
Description
Emotion recognition in conversation has applications within numerous domains such as affective computing and medicine. Recent methods for emotion recognition jointly utilize conversational data over several modalities including audio, video, and text. However, state-of-the-art frameworks for this task do not focus on the feature extraction and feature fusion steps of this process. This thesis aims to improve the state-of-the-art method by incorporating two components to better accomplish these steps. By doing so, we are able to produce improved representations for the text modality and better model the relationships between all modalities. This paper proposes two methods which focus on these concepts and provide improved accuracy over the state-of-the-art framework for multimodal emotion recognition in dialogue.
ContributorsRawal, Siddharth (Author) / Baral, Chitta (Thesis director) / Shah, Shrikant (Committee member) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2020-05
Description
In recent years, the development of new Machine Learning models has allowed for new technological advancements to be introduced for practical use across the world. Multiple studies and experiments have been conducted to create new variations of Machine Learning models with different algorithms to determine if potential systems would prove to be successful. Even today, there are still many research initiatives that are continuing to develop new models in the hopes to discover potential solutions for problems such as autonomous driving or determining the emotional value from a single sentence. One of the current popular research topics for Machine Learning is the development of Facial Expression Recognition systems. These Machine Learning models focus on classifying images of human faces that are expressing different emotions through facial expressions. In order to develop effective models to perform Facial Expression Recognition, researchers have gone on to utilize Deep Learning models, which are a more advanced implementation of Machine Learning models, known as Neural Networks. More specifically, the use of Convolutional Neural Networks has proven to be the most effective models for achieving highly accurate results at classifying images of various facial expressions. Convolutional Neural Networks are Deep Learning models that are capable of processing visual data, such as images and videos, and can be used to identify various facial expressions. The purpose of this project, I focused on learning about the important concepts of Machine Learning, Deep Learning, and Convolutional Neural Networks to implement a Convolutional Neural Network that was previously developed by a recommended research paper.
ContributorsFrace, Douglas R (Author) / Demakethepalli Venkateswara, Hemanth Kumar (Thesis director) / McDaniel, Troy (Committee member) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2020-05
Description
Abstract
Matrix Factorization techniques have been proven to be more effective in recommender systems than standard user based or item based methods. Using this knowledge, Funk SVD and SVD++ are compared by the accuracy of their predictions of Twitch streamer data.
Introduction
As watching video games is becoming more popular, those interested are becoming interested in Twitch.tv, an online platform for guests to watch streamers play video games and interact with them. A streamer is an person who broadcasts them-self playing a video game or some other thing for an audience (the guests of the website.) The site allows the guest to first select the game/category to view and then displays currently active streamers for the guest to select and watch. Twitch records the games that a streamer plays along with the amount of time that a streamer spends streaming that game. This is how the score is generated for a streamer’s game. These three terms form the streamer-game-score (user-item-rating) tuples that we use to train out models.
The our problem’s solution is similar to the purpose of the Netflix prize; however, as opposed to suggesting a user a movie, the goal is to suggest a user a game. We built a model to predict the score that a streamer will have for a game. The score field in our data is fundamentally different from a movie rating in Netflix because the way a user influences a game’s score is by actively streaming it, not by giving it an score based off opinion. The dataset being used it the Twitch.tv dataset provided by Isaac Jones [1]. Also, the only data used in training the models is in the form of the streamer-game-score (user-item-rating) tuples. It will be known if these data points with limited information will be able to give an accurate prediction of a streamer’s score for a game. SVD and SVD++ are the baseis of the models being trained and tested. Scikit’s Surprise library in Python3 is used for the implementation of the models.
Matrix Factorization techniques have been proven to be more effective in recommender systems than standard user based or item based methods. Using this knowledge, Funk SVD and SVD++ are compared by the accuracy of their predictions of Twitch streamer data.
Introduction
As watching video games is becoming more popular, those interested are becoming interested in Twitch.tv, an online platform for guests to watch streamers play video games and interact with them. A streamer is an person who broadcasts them-self playing a video game or some other thing for an audience (the guests of the website.) The site allows the guest to first select the game/category to view and then displays currently active streamers for the guest to select and watch. Twitch records the games that a streamer plays along with the amount of time that a streamer spends streaming that game. This is how the score is generated for a streamer’s game. These three terms form the streamer-game-score (user-item-rating) tuples that we use to train out models.
The our problem’s solution is similar to the purpose of the Netflix prize; however, as opposed to suggesting a user a movie, the goal is to suggest a user a game. We built a model to predict the score that a streamer will have for a game. The score field in our data is fundamentally different from a movie rating in Netflix because the way a user influences a game’s score is by actively streaming it, not by giving it an score based off opinion. The dataset being used it the Twitch.tv dataset provided by Isaac Jones [1]. Also, the only data used in training the models is in the form of the streamer-game-score (user-item-rating) tuples. It will be known if these data points with limited information will be able to give an accurate prediction of a streamer’s score for a game. SVD and SVD++ are the baseis of the models being trained and tested. Scikit’s Surprise library in Python3 is used for the implementation of the models.
ContributorsAitken, Connor Dalton (Author) / Liu, Huan (Thesis director) / Jones, Isaac (Committee member) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2019-05
Description
Propaganda bots are malicious bots on Twitter that spread divisive opinions and support political accounts. This project is based on detecting propaganda bots on Twitter using machine learning. Once I began to observe patterns within propaganda followers on Twitter, I determined that I could train algorithms to detect these bots. The paper focuses on my development and process of training classifiers and using them to create a user-facing server that performs prediction functions automatically. The learning goals of this project were detailed, the focus of which was to learn some form of machine learning architecture. I needed to learn some aspect of large data handling, as well as being able to maintain these datasets for training use. I also needed to develop a server that would execute these functionalities on command. I wanted to be able to design a full-stack system that allowed me to create every aspect of a user-facing server that can execute predictions using the classifiers that I design.
Throughout this project, I decided on a number of learning goals to consider it a success. I needed to learn how to use the supporting libraries that would help me to design this system. I also learned how to use the Twitter API, as well as create the infrastructure behind it that would allow me to collect large amounts of data for machine learning. I needed to become familiar with common machine learning libraries in Python in order to create the necessary algorithms and pipelines to make predictions based on Twitter data.
This paper details the steps and decisions needed to determine how to collect this data and apply it to machine learning algorithms. I determined how to create labelled data using pre-existing Botometer ratings, and the levels of confidence I needed to label data for training. I use the scikit-learn library to create these algorithms to best detect these bots. I used a number of pre-processing routines to refine the classifiers’ precision, including natural language processing and data analysis techniques. I eventually move to remotely-hosted versions of the system on Amazon web instances to collect larger amounts of data and train more advanced classifiers. This leads to the details of my final implementation of a user-facing server, hosted on AWS and interfacing over Gmail’s IMAP server.
The current and future development of this system is laid out. This includes more advanced classifiers, better data analysis, conversions to third party Twitter data collection systems, and user features. I detail what it is I have learned from this exercise, and what it is I hope to continue working on.
Throughout this project, I decided on a number of learning goals to consider it a success. I needed to learn how to use the supporting libraries that would help me to design this system. I also learned how to use the Twitter API, as well as create the infrastructure behind it that would allow me to collect large amounts of data for machine learning. I needed to become familiar with common machine learning libraries in Python in order to create the necessary algorithms and pipelines to make predictions based on Twitter data.
This paper details the steps and decisions needed to determine how to collect this data and apply it to machine learning algorithms. I determined how to create labelled data using pre-existing Botometer ratings, and the levels of confidence I needed to label data for training. I use the scikit-learn library to create these algorithms to best detect these bots. I used a number of pre-processing routines to refine the classifiers’ precision, including natural language processing and data analysis techniques. I eventually move to remotely-hosted versions of the system on Amazon web instances to collect larger amounts of data and train more advanced classifiers. This leads to the details of my final implementation of a user-facing server, hosted on AWS and interfacing over Gmail’s IMAP server.
The current and future development of this system is laid out. This includes more advanced classifiers, better data analysis, conversions to third party Twitter data collection systems, and user features. I detail what it is I have learned from this exercise, and what it is I hope to continue working on.
ContributorsPeterson, Austin (Author) / Yang, Yezhou (Thesis director) / Sadasivam, Aadhavan (Committee member) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2019-05
Description
Machine learning is a powerful tool for processing and understanding the vast amounts of data produced by sensors every day. Machine learning has found use in a wide variety of fields, from making medical predictions through correlations invisible to the human eye to classifying images in computer vision applications. A wide range of machine learning algorithms have been developed to attempt to solve these problems, each with different metrics in accuracy, throughput, and energy efficiency. However, even after they are trained, these algorithms require substantial computations to make a prediction. General-purpose CPUs are not well-optimized to this task, so other hardware solutions have developed over time, including the use of a GPU, FPGA, or ASIC.
This project considers the FPGA implementations of MLP and CNN feedforward. While FPGAs provide significant performance improvements, they come at a substantial financial cost. We explore the options of implementing these algorithms on a smaller budget. We successfully implement a multilayer perceptron that identifies handwritten digits from the MNIST dataset on a student-level DE10-Lite FPGA with a test accuracy of 91.99%. We also apply our trained network to external image data loaded through a webcam and a Raspberry Pi, but we observe lower test accuracy in these images. Later, we consider the requirements necessary to implement a more elaborate convolutional neural network on the same FPGA. The study deems the CNN implementation feasible in the criteria of memory requirements and basic architecture. We suggest the CNN implementation on the same FPGA to be worthy of further exploration.
This project considers the FPGA implementations of MLP and CNN feedforward. While FPGAs provide significant performance improvements, they come at a substantial financial cost. We explore the options of implementing these algorithms on a smaller budget. We successfully implement a multilayer perceptron that identifies handwritten digits from the MNIST dataset on a student-level DE10-Lite FPGA with a test accuracy of 91.99%. We also apply our trained network to external image data loaded through a webcam and a Raspberry Pi, but we observe lower test accuracy in these images. Later, we consider the requirements necessary to implement a more elaborate convolutional neural network on the same FPGA. The study deems the CNN implementation feasible in the criteria of memory requirements and basic architecture. We suggest the CNN implementation on the same FPGA to be worthy of further exploration.
ContributorsLythgoe, Zachary James (Author) / Allee, David (Thesis director) / Hartin, Olin (Committee member) / Electrical Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2019-12
Description
Vulnerability testing/evaluation is a regular task for cyber-security groups. Conducting tasks like this can take up a great amount of time and may not be perfect. Automating these tasks helps speed up the rate at which experts can test systems. However, script based or static programs that run automatically often do not have the versatility required to properly replace human analysis. With the advances in Artificial Intelligence and Machine Learning, a utility can be developed that would allow for the creation of penetration testing plans rather than manually testing vulnerabilities. A variety of existing cyber-security programs and utilities provide an API layer that commonly interacts with the Python environment. With the commonality of AI/ML tools within the Python ecosystem, a plugin like interface can be developed to feed any AI/ML program real world data and receive a response/report in return. Using Python 2.7+, Python 3.6+, pymdptoolbox, and POMDPy, a program was developed that ingests real-world data from scanning tools and returned a suggested course of action to be used by analysts in order to perform a practical validation of the algorithms in a real world setting. This program was able to successfully navigate a test network and produce results that were expected to be found on the target machines without needing human analysis of the network. Using POMDP based systems for more cyber-security type tasks may be a valuable use case for future developments and help ease the burden faced in a rapid paced world.
ContributorsBelanger, Connor Lawrence (Author) / Huang, Dijiang (Thesis director) / Chowdhary, Ankur (Committee member) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2020-05
Description
Multi-view learning, a subfield of machine learning that aims to improve model performance by training on multiple views of the data, has been studied extensively in the past decades. It is typically applied in contexts where the input features naturally form multiple groups or views. An example of a naturally multi-view context is a data set of websites, where each website is described not only by the text on the page, but also by the text of hyperlinks pointing to the page. More recently, various studies have demonstrated the initial success of applying multi-view learning on single-view data with multiple artificially constructed views. However, there lacks a systematic study regarding the effectiveness of such artificially constructed views. To bridge this gap, this thesis begins by providing a high-level overview of multi-view learning with the co-training algorithm. Co-training is a classic semi-supervised learning algorithm that takes advantage of both labelled and unlabelled examples in the data set for training. Then, the thesis presents a web-based tool developed in Python allowing users to experiment with and compare the performance of multiple view construction approaches on various data sets. The supported view construction approaches in the web-based tool include subsampling, Optimal Feature Set Partitioning, and the genetic algorithm. Finally, the thesis presents an empirical comparison of the performance of these approaches, not only against one another, but also against traditional single-view models. The findings show that a simple subsampling approach combined with co-training often outperforms both the other view construction approaches, as well as traditional single-view methods.
ContributorsAksoy, Kaan (Author) / Maciejewski, Ross (Thesis director) / He, Jingrui (Committee member) / Computer Science and Engineering Program (Contributor) / Barrett, The Honors College (Contributor)
Created2019-12