Current Search: Machine Learning (x)
View All Items
(1 - 20 of 60)
Pages
- Title
- Robust, Scalable, and Provable Approaches to High Dimensional Unsupervised Learning.
- Creator
-
Rahmani, Mostafa, Atia, George, Vosoughi, Azadeh, Mikhael, Wasfy, Nashed, M, Pensky, Marianna, University of Central Florida
- Abstract / Description
-
This doctoral thesis focuses on three popular unsupervised learning problems: subspace clustering, robust PCA, and column sampling. For the subspace clustering problem, a new transformative idea is presented. The proposed approach, termed Innovation Pursuit, is a new geometrical solution to the subspace clustering problem whereby subspaces are identified based on their relative novelties. A detailed mathematical analysis is provided establishing sufficient conditions for the proposed method...
Show moreThis doctoral thesis focuses on three popular unsupervised learning problems: subspace clustering, robust PCA, and column sampling. For the subspace clustering problem, a new transformative idea is presented. The proposed approach, termed Innovation Pursuit, is a new geometrical solution to the subspace clustering problem whereby subspaces are identified based on their relative novelties. A detailed mathematical analysis is provided establishing sufficient conditions for the proposed method to correctly cluster the data points. The numerical simulations with both real and synthetic data demonstrate that Innovation Pursuit notably outperforms the state-of-the-art subspace clustering algorithms. For the robust PCA problem, we focus on both the outlier detection and the matrix decomposition problems. For the outlier detection problem, we present a new algorithm, termed Coherence Pursuit, in addition to two scalable randomized frameworks for the implementation of outlier detection algorithms. The Coherence Pursuit method is the first provable and non-iterative robust PCA method which is provably robust to both unstructured and structured outliers. Coherence Pursuit is remarkably simple and it notably outperforms the existing methods in dealing with structured outliers. In the proposed randomized designs, we leverage the low dimensional structure of the low rank component to apply the robust PCA algorithm to a random sketch of the data as opposed to the full scale data. Importantly, it is analytically shown that the presented randomized designs can make the computation or sample complexity of the low rank matrix recovery algorithm independent of the size of the data. At the end, we focus on the column sampling problem. A new sampling tool, dubbed Spatial Random Sampling, is presented which performs the random sampling in the spatial domain. The most compelling feature of Spatial Random Sampling is that it is the first unsupervised column sampling method which preserves the spatial distribution of the data.
Show less - Date Issued
- 2018
- Identifier
- CFE0007083, ucf:52010
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0007083
- Title
- Training Neural Networks Through the Integration of Evolution and Gradient Descent.
- Creator
-
Morse, Gregory, Stanley, Kenneth, Wu, Annie, Shah, Mubarak, Wiegand, Rudolf, University of Central Florida
- Abstract / Description
-
Neural networks have achieved widespread adoption due to both their applicability to a wide range of problems and their success relative to other machine learning algorithms. The training of neural networks is achieved through any of several paradigms, most prominently gradient-based approaches (including deep learning), but also through up-and-coming approaches like neuroevolution. However, while both of these neural network training paradigms have seen major improvements over the past...
Show moreNeural networks have achieved widespread adoption due to both their applicability to a wide range of problems and their success relative to other machine learning algorithms. The training of neural networks is achieved through any of several paradigms, most prominently gradient-based approaches (including deep learning), but also through up-and-coming approaches like neuroevolution. However, while both of these neural network training paradigms have seen major improvements over the past decade, little work has been invested in developing algorithms that incorporate the advances from both deep learning and neuroevolution. This dissertation introduces two new algorithms that are steps towards the integration of gradient descent and neuroevolution for training neural networks. The first is (1) the Limited Evaluation Evolutionary Algorithm (LEEA), which implements a novel form of evolution where individuals are partially evaluated, allowing rapid learning and enabling the evolutionary algorithm to behave more like gradient descent. This conception provides a critical stepping stone to future algorithms that more tightly couple evolutionary and gradient descent components. The second major algorithm (2) is Divergent Discriminative Feature Accumulation (DDFA), which combines a neuroevolution phase, where features are collected in an unsupervised manner, with a gradient descent phase for fine tuning of the neural network weights. The neuroevolution phase of DDFA utilizes an indirect encoding and novelty search, which are sophisticated neuroevolution components rarely incorporated into gradient descent-based systems. Further contributions of this work that build on DDFA include (3) an empirical analysis to identify an effective distance function for novelty search in high dimensions and (4) the extension of DDFA for the purpose of discovering convolutional features. The results of these DDFA experiments together show that DDFA discovers features that are effective as a starting point for gradient descent, with significant improvement over gradient descent alone. Additionally, the method of collecting features in an unsupervised manner allows DDFA to be applied to domains with abundant unlabeled data and relatively sparse labeled data. This ability is highlighted in the STL-10 domain, where DDFA is shown to make effective use of unlabeled data.
Show less - Date Issued
- 2019
- Identifier
- CFE0007840, ucf:52819
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0007840
- Title
- A STRUCTURAL AND FUNCTIONAL ANALYSIS OF HUMAN BRAIN MRI WITH ATTENTION DEFICIT HYPERACTIVITY DISORDER.
- Creator
-
Watane, Arjun A, Bagci, Ulas, University of Central Florida
- Abstract / Description
-
Attention Deficit Hyperactivity Disorder (ADHD) affects 5-10% of children worldwide. Its effects are mainly behavioral, manifesting in symptoms such as inattention, hyperactivity, and impulsivity. If not monitored and treated, ADHD may adversely affect a child's health, education, and social life. Furthermore, the neurological disorder is currently diagnosed through interviews and opinions of teachers, parents, and physicians. Because this is a subjective method of identifying ADHD, it is...
Show moreAttention Deficit Hyperactivity Disorder (ADHD) affects 5-10% of children worldwide. Its effects are mainly behavioral, manifesting in symptoms such as inattention, hyperactivity, and impulsivity. If not monitored and treated, ADHD may adversely affect a child's health, education, and social life. Furthermore, the neurological disorder is currently diagnosed through interviews and opinions of teachers, parents, and physicians. Because this is a subjective method of identifying ADHD, it is easily prone to error and misdiagnosis. Therefore, there is a clear need to develop an objective diagnostic method for ADHD. The focus of this study is to explore the use of machine language classifiers on information from the brain MRI and fMRI of both ADHD and non-ADHD subjects. The imaging data are preprocessed to remove any intra-subject and inter-subject variation. For both MRI and fMRI, similar preprocessing stages are performed, including normalization, skull stripping, realignment, smoothing, and co-registration. The next step is to extract features from the data. For MRI, anatomical features such as cortical thickness, surface area, volume, and intensity are obtained. For fMRI, region of interest (ROI) correlation coefficients between 116 cortical structures are determined. A large number of image features are collected, yet many of them may include redundant and useless information. Therefore, the features used for training and testing the classifiers are selected in two separate ways, feature ranking and stability selection, and their results are compared. Once the best features from MRI and fMRI are determined, the following classifiers are trained and tested through leave-one-out cross validation, experimenting with varying feature numbers, for each imaging modality and feature selection method: support vector machine, support vector regression, random forest, and elastic net. Thus, there are four experiments (MRI-rank, MRI-stability, fMRI-rank, fMRI-stability) with four classifiers in each for a total of 16 classifiers trained per each feature count attempted. The results of each classifier are the decisions of each subject, ADHD or non-ADHD. Finally, a classifier decision ensemble is created through the combination of the outputs of the best classifiers in a majority voting method that includes results of both the MRI and fMRI classifiers and keeps both feature selection results independent. The results suggest that ADHD is more easily identified through fMRI because the classification accuracies are a lot higher using fMRI data rather than MRI data. Furthermore, significant activity correlation differences exist between the brain's frontal lobe and cerebellum and also the left and right hemispheres among ADHD and non-ADHD subjects. When including MRI decisions with fMRI in the classifier ensemble, performance is boosted to a high ADHD detection accuracy of 96.2%, suggesting that MRI information assists in validating fMRI classification decisions. This study is an important step towards the development of an automatic and objective method for ADHD diagnosis. While more work is needed to externally validate and improve the classification accuracy, new applications of current methods with promising results are introduced here.
Show less - Date Issued
- 2017
- Identifier
- CFH2000203, ucf:45978
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFH2000203
- Title
- Relating First-person and Third-person Vision.
- Creator
-
Ardeshir Behrostaghi, Shervin, Borji, Ali, Shah, Mubarak, Hu, Haiyan, Atia, George, University of Central Florida
- Abstract / Description
-
Thanks to the availability and increasing popularity of wearable devices such as GoPro cameras, smart phones and glasses, we have access to a plethora of videos captured from the first person (egocentric) perspective. Capturing the world from the perspective of one's self, egocentric videos bear characteristics distinct from the more traditional third-person (exocentric) videos. In many computer vision tasks (e.g. identification, action recognition, face recognition, pose estimation, etc.),...
Show moreThanks to the availability and increasing popularity of wearable devices such as GoPro cameras, smart phones and glasses, we have access to a plethora of videos captured from the first person (egocentric) perspective. Capturing the world from the perspective of one's self, egocentric videos bear characteristics distinct from the more traditional third-person (exocentric) videos. In many computer vision tasks (e.g. identification, action recognition, face recognition, pose estimation, etc.), the human actors are the main focus. Hence, detecting, localizing, and recognizing the human actor is often incorporated as a vital component. In an egocentric video however, the person behind the camera is often the person of interest. This would change the nature of the task at hand, given that the camera holder is usually not visible in the content of his/her egocentric video. In other words, our knowledge about the visual appearance, pose, etc. on the egocentric camera holder is very limited, suggesting reliance on other cues in first person videos. First and third person videos have been separately studied in the past in the computer vision community. However, the relationship between first and third person vision has yet to be fully explored. Relating these two views systematically could potentially benefit many computer vision tasks and applications. This thesis studies this relationship in several aspects. We explore supervised and unsupervised approaches for relating these two views seeking different objectives such as identification, temporal alignment, and action classification. We believe that this exploration could lead to a better understanding the relationship of these two drastically different sources of information.
Show less - Date Issued
- 2018
- Identifier
- CFE0007151, ucf:52322
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0007151
- Title
- Methods for online feature selection for classification problems.
- Creator
-
Razmjoo, Alaleh, Zheng, Qipeng, Rabelo, Luis, Boginski, Vladimir, Xanthopoulos, Petros, University of Central Florida
- Abstract / Description
-
Online learning is a growing branch of machine learning which allows all traditional data miningtechniques to be applied on an online stream of data in real-time. In this dissertation, we presentthree efficient algorithms for feature ranking in online classification problems. Each of the methodsare tailored to work well with different types of classification tasks and have different advantages.The reason for this variety of algorithms is that like other machine learning solutions, there is...
Show moreOnline learning is a growing branch of machine learning which allows all traditional data miningtechniques to be applied on an online stream of data in real-time. In this dissertation, we presentthree efficient algorithms for feature ranking in online classification problems. Each of the methodsare tailored to work well with different types of classification tasks and have different advantages.The reason for this variety of algorithms is that like other machine learning solutions, there is usuallyno algorithm which works well for all types of tasks. The first method, is an online sensitivitybased feature ranking (SFR) which is updated incrementally, and is designed for classificationtasks with continuous features. We take advantage of the concept of global sensitivity and rankfeatures based on their impact on the outcome of the classification model. In the feature selectionpart, we use a two-stage filtering method in order to first eliminate highly correlated and redundantfeatures and then eliminate irrelevant features in the second stage. One important advantage of ouralgorithm is its generality, which means the method works for correlated feature spaces withoutpreprocessing. It can be implemented along with any single-pass online classification method withseparating hyperplane such as SVMs. In the second method, with help of probability theory wepropose an algorithm which measures the importance of the features by observing the changes inlabel prediction in case of feature substitution. A non-parametric version of the proposed methodis presented to eliminate the distribution type assumptions. These methods are application to alldata types including mixed feature spaces. At last, we present a class-based feature importanceranking method which evaluates the importance of each feature for each class, these sub-rankingsare further exploited to train an ensemble of classifiers. The proposed methods will be thoroughlytested using benchmark datasets and the results will be discussed in the last chapter.
Show less - Date Issued
- 2018
- Identifier
- CFE0007584, ucf:52567
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0007584
- Title
- D-FENS: DNS Filtering (&) Extraction Network System for Malicious Domain Names.
- Creator
-
Spaulding, Jeffrey, Mohaisen, Aziz, Leavens, Gary, Bassiouni, Mostafa, Fu, Xinwen, Posey, Clay, University of Central Florida
- Abstract / Description
-
While the DNS (Domain Name System) has become a cornerstone for the operation of the Internet, it has also fostered creative cases of maliciousness, including phishing, typosquatting, and botnet communication among others. To address this problem, this dissertation focuses on identifying and mitigating such malicious domain names through prior knowledge and machine learning. In the first part of this dissertation, we explore a method of registering domain names with deliberate typographical...
Show moreWhile the DNS (Domain Name System) has become a cornerstone for the operation of the Internet, it has also fostered creative cases of maliciousness, including phishing, typosquatting, and botnet communication among others. To address this problem, this dissertation focuses on identifying and mitigating such malicious domain names through prior knowledge and machine learning. In the first part of this dissertation, we explore a method of registering domain names with deliberate typographical mistakes (i.e., typosquatting) to masquerade as popular and well-established domain names. To understand the effectiveness of typosquatting, we conducted a user study which helped shed light on which techniques were more (")successful(") than others in deceiving users. While certain techniques fared better than others, they failed to take the context of the user into account. Therefore, in the second part of this dissertation we look at the possibility of an advanced attack which takes context into account when generating domain names. The main idea is determining the possibility for an adversary to improve their (")success(") rate of deceiving users with specifically-targeted malicious domain names. While these malicious domains typically target users, other types of domain names are generated by botnets for command (&) control (C2) communication. Therefore, in the third part of this dissertation we investigate domain generation algorithms (DGA) used by botnets and propose a method to identify DGA-based domain names. By analyzing DNS traffic for certain patterns of NXDomain (non-existent domain) query responses, we can accurately predict DGA-based domain names before they are registered. Given all of these approaches to malicious domain names, we ultimately propose a system called D-FENS (DNS Filtering (&) Extraction Network System). D-FENS uses machine learning and prior knowledge to accurately predict unreported malicious domain names in real-time, thereby preventing Internet devices from unknowingly connecting to a potentially malicious domain name.
Show less - Date Issued
- 2018
- Identifier
- CFE0007587, ucf:52540
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0007587
- Title
- A CONTEXTUAL APPROACH TO LEARNING COLLABORATIVE BEHAVIOR VIA OBSERVATION.
- Creator
-
Johnson, Cynthia, Gonzalez, Avelino, University of Central Florida
- Abstract / Description
-
This dissertation describes a novel technique to creating a simulated team of agents through observation. Simulated human teamwork can be used for a number of purposes, such as expert examples, automated teammates for training purposes and realistic opponents in games and training simulation. Current teamwork simulations require the team member behaviors be programmed into the simulation, often requiring a great deal of time and effort. None are able to observe a team at work and replicate...
Show moreThis dissertation describes a novel technique to creating a simulated team of agents through observation. Simulated human teamwork can be used for a number of purposes, such as expert examples, automated teammates for training purposes and realistic opponents in games and training simulation. Current teamwork simulations require the team member behaviors be programmed into the simulation, often requiring a great deal of time and effort. None are able to observe a team at work and replicate the teamwork behaviors. Machine learning techniques for learning by observation and learning by demonstration have proven successful at observing behavior of humans or other software agents and creating a behavior function for a single agent. The research described here combines current research in teamwork simulations and learning by observation to effectively train a multi-agent system in effective team behavior. The dissertation describes the background and work by others as well as a detailed description of the learning method. A prototype built to evaluate the developed approach as well as the extensive experimentation conducted is also described.
Show less - Date Issued
- 2011
- Identifier
- CFE0003602, ucf:48869
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0003602
- Title
- Machine Learning from Casual Conversation.
- Creator
-
Mohammed Ali, Awrad, Sukthankar, Gita, Wu, Annie, Boloni, Ladislau, University of Central Florida
- Abstract / Description
-
Human social learning is an effective process that has inspired many existing machine learning techniques, such as learning from observation and learning by demonstration. In this dissertation, we introduce another form of social learning, Learning from a Casual Conversation (LCC). LCC is an open-ended machine learning system in which an artificially intelligent agent learns from an extended dialog with a human. Our system enables the agent to incorporate changes into its knowledge base,...
Show moreHuman social learning is an effective process that has inspired many existing machine learning techniques, such as learning from observation and learning by demonstration. In this dissertation, we introduce another form of social learning, Learning from a Casual Conversation (LCC). LCC is an open-ended machine learning system in which an artificially intelligent agent learns from an extended dialog with a human. Our system enables the agent to incorporate changes into its knowledge base, based on the human's conversational text input. This system emulates how humans learn from each other through a dialog. LCC closes the gap in the current research that is focused on teaching specific tasks to computer agents. Furthermore, LCC aims to provide an easy way to enhance the knowledge of the system without requiring the involvement of a programmer. This system does not require the user to enter specific information; instead, the user can chat naturally with the agent. LCC identifies the inputs that contain information relevant to its knowledge base in the learning process. LCC's architecture consists of multiple sub-systems combined to perform the task. Its learning component can add new knowledge to existing information in the knowledge base, confirm existing information, and/or update existing information found to be related to the user input. %The test results indicate that the prototype was successful in learning from a conversation. The LCC system functionality was assessed using different evaluation methods. This includes tests performed by the developer, as well as by 130 human test subjects. Thirty of those test subjects interacted directly with the system and completed a survey of 13 questions/statements that asked the user about his/her experience using LCC. A second group of 100 human test subjects evaluated the dialogue logs of a subset of the first group of human testers. The collected results were all found to be acceptable and within the range of our expectations.
Show less - Date Issued
- 2019
- Identifier
- CFE0007503, ucf:52634
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0007503
- Title
- On Kernel-base Multi-Task Learning.
- Creator
-
Li, Cong, Georgiopoulos, Michael, Anagnostopoulos, Georgios, Tappen, Marshall, Hu, Haiyan, Ni, Liqiang, University of Central Florida
- Abstract / Description
-
Multi-Task Learning (MTL) has been an active research area in machine learning for two decades. By training multiple relevant tasks simultaneously with information shared across tasks, it is possible to improve the generalization performance of each task, compared to training each individual task independently. During the past decade, most MTL research has been based on the Regularization-Loss framework due to its flexibility in specifying various types of information sharing strategies, the...
Show moreMulti-Task Learning (MTL) has been an active research area in machine learning for two decades. By training multiple relevant tasks simultaneously with information shared across tasks, it is possible to improve the generalization performance of each task, compared to training each individual task independently. During the past decade, most MTL research has been based on the Regularization-Loss framework due to its flexibility in specifying various types of information sharing strategies, the opportunity it offers to yield a kernel-based methods and its capability in promoting sparse feature representations.However, certain limitations exist in both theoretical and practical aspects of Regularization-Loss-based MTL. Theoretically, previous research on generalization bounds in connection to MTL Hypothesis Space (HS)s, where data of all tasks are pre-processed by a (partially) common operator, has been limited in two aspects: First, all previous works assumed linearity of the operator, therefore completely excluding kernel-based MTL HSs, for which the operator is potentially non-linear. Secondly, all previous works, rather unnecessarily, assumed that all the task weights to be constrained within norm-balls, whose radii are equal. The requirement of equal radii leads to significant inflexibility of the relevant HSs, which may cause the generalization performance of the corresponding MTL models to deteriorate. Practically, various algorithms have been developed for kernel-based MTL models, due to different characteristics of the formulations. Most of these algorithms are a burden to develop and end up being quite sophisticated, so that practitioners may face a hard task in interpreting and implementing them, especially when multiple models are involved. This is even more so, when Multi-Task Multiple Kernel Learning (MT-MKL) models are considered. This research largely resolves the above limitations. Theoretically, a pair of new kernel-based HSs are proposed: one for single-kernel MTL, and another one for MT-MKL. Unlike previous works, we allow each task weight to be constrained within a norm-ball, whose radius is learned during training. By deriving and analyzing the generalization bounds of these two HSs, we show that, indeed, such a flexibility leads to much tighter generalization bounds, which often results to significantly better generalization performance. Based on this observation, a pair of new models is developed, one for each case: single-kernel MTL, and another one for MT-MKL. From a practical perspective, we propose a general MT-MKL framework that covers most of the prominent MT-MKL approaches, including our new MT-MKL formulation. Then, a general purpose algorithm is developed to solve the framework, which can also be employed for training all other models subsumed by this framework. A series of experiments is conducted to assess the merits of the proposed mode when trained by the new algorithm. Certain properties of our HSs and formulations are demonstrated, and the advantage of our model in terms of classification accuracy is shown via these experiments.
Show less - Date Issued
- 2014
- Identifier
- CFE0005517, ucf:50321
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0005517
- Title
- Life Long Learning in Sparse Learning Environments.
- Creator
-
Reeder, John, Georgiopoulos, Michael, Gonzalez, Avelino, Sukthankar, Gita, Anagnostopoulos, Georgios, University of Central Florida
- Abstract / Description
-
Life long learning is a machine learning technique that deals with learning sequential tasks over time. It seeks to transfer knowledge from previous learning tasks to new learning tasks in order to increase generalization performance and learning speed. Real-time learning environments in which many agents are participating may provide learning opportunities but they are spread out in time and space outside of the geographical scope of a single learning agent. This research seeks to provide an...
Show moreLife long learning is a machine learning technique that deals with learning sequential tasks over time. It seeks to transfer knowledge from previous learning tasks to new learning tasks in order to increase generalization performance and learning speed. Real-time learning environments in which many agents are participating may provide learning opportunities but they are spread out in time and space outside of the geographical scope of a single learning agent. This research seeks to provide an algorithm and framework for life long learning among a network of agents in a sparse real-time learning environment. This work will utilize the robust knowledge representation of neural networks, and make use of both functional and representational knowledge transfer to accomplish this task. A new generative life long learning algorithm utilizing cascade correlation and reverberating pseudo-rehearsal and incorporating a method for merging divergent life long learning paths will be implemented.
Show less - Date Issued
- 2013
- Identifier
- CFE0004917, ucf:49601
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0004917
- Title
- Complementary Layered Learning.
- Creator
-
Mondesire, Sean, Wu, Annie, Wiegand, Rudolf, Sukthankar, Gita, Proctor, Michael, University of Central Florida
- Abstract / Description
-
Layered learning is a machine learning paradigm used to develop autonomous robotic-based agents by decomposing a complex task into simpler subtasks and learns each sequentially. Although the paradigm continues to have success in multiple domains, performance can be unexpectedly unsatisfactory. Using Boolean-logic problems and autonomous agent navigation, we show poor performance is due to the learner forgetting how to perform earlier learned subtasks too quickly (favoring plasticity) or...
Show moreLayered learning is a machine learning paradigm used to develop autonomous robotic-based agents by decomposing a complex task into simpler subtasks and learns each sequentially. Although the paradigm continues to have success in multiple domains, performance can be unexpectedly unsatisfactory. Using Boolean-logic problems and autonomous agent navigation, we show poor performance is due to the learner forgetting how to perform earlier learned subtasks too quickly (favoring plasticity) or having difficulty learning new things (favoring stability). We demonstrate that this imbalance can hinder learning so that task performance is no better than that of a sub-optimal learning technique, monolithic learning, which does not use decomposition. Through the resulting analyses, we have identified factors that can lead to imbalance and their negative effects, providing a deeper understanding of stability and plasticity in decomposition-based approaches, such as layered learning.To combat the negative effects of the imbalance, a complementary learning system is applied to layered learning. The new technique augments the original learning approach with dual storage region policies to preserve useful information from being removed from an agent's policy prematurely. Through multi-agent experiments, a 28% task performance increase is obtained with the proposed augmentations over the original technique.
Show less - Date Issued
- 2014
- Identifier
- CFE0005213, ucf:50626
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0005213
- Title
- Learning Internal State Memory Representations from Observation.
- Creator
-
Wong, Josiah, Gonzalez, Avelino, Liu, Fei, Wu, Annie, Ontanon, Santiago, Wiegand, Rudolf, University of Central Florida
- Abstract / Description
-
Learning from Observation (LfO) is a machine learning paradigm that mimics how people learn in daily life: learning how to do something simply by watching someone else do it. LfO has been used in various applications, from video game agent creation to driving a car, but it has always been limited by the inability of an observer to know what a performing entity chooses to remember as they act in an environment. Various methods have either ignored the effects of memory or otherwise made...
Show moreLearning from Observation (LfO) is a machine learning paradigm that mimics how people learn in daily life: learning how to do something simply by watching someone else do it. LfO has been used in various applications, from video game agent creation to driving a car, but it has always been limited by the inability of an observer to know what a performing entity chooses to remember as they act in an environment. Various methods have either ignored the effects of memory or otherwise made simplistic assumptions about its structure. In this dissertation, we propose a new method, Memory Composition Learning, that captures the influence of a performer's memory in an observed behavior through the creation of an auxiliary memory feature set that explicitly models the aspects of the environment with significance for future decisions, and which can be used with a machine learning technique to provide salient information from memory. It advances the state of the art by automatically learning the internal structure of memory instead of ignoring or predefining it. This research is difficult in that memory modeling is an unsupervised learning problem that we elect to solve solely from unobtrusive observation. This research is significant for LfO in that it will allow learning techniques that otherwise could not use information from memory to use a tailored set of learned memory features that capture salient influences from memory and enable decision-making based on these influences for more effective learning performance. To validate our hypothesis, we implemented a prototype for modeling observed memory influences with our approach and applied it to simulated vacuum cleaner and lawn mower domains. Our investigation revealed that MCL was able to automatically learn memory features that describe the influences on an observed actor's internal state, and which improved learning performance of observed behaviors.
Show less - Date Issued
- 2019
- Identifier
- CFE0007879, ucf:52755
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0007879
- Title
- Online, Supervised and Unsupervised Action Localization in Videos.
- Creator
-
Soomro, Khurram, Shah, Mubarak, Heinrich, Mark, Hu, Haiyan, Bagci, Ulas, Yun, Hae-Bum, University of Central Florida
- Abstract / Description
-
Action recognition classifies a given video among a set of action labels, whereas action localization determines the location of an action in addition to its class. The overall aim of this dissertation is action localization. Many of the existing action localization approaches exhaustively search (spatially and temporally) for an action in a video. However, as the search space increases with high resolution and longer duration videos, it becomes impractical to use such sliding window...
Show moreAction recognition classifies a given video among a set of action labels, whereas action localization determines the location of an action in addition to its class. The overall aim of this dissertation is action localization. Many of the existing action localization approaches exhaustively search (spatially and temporally) for an action in a video. However, as the search space increases with high resolution and longer duration videos, it becomes impractical to use such sliding window techniques. The first part of this dissertation presents an efficient approach for localizing actions by learning contextual relations between different video regions in training. In testing, we use the context information to estimate the probability of each supervoxel belonging to the foreground action and use Conditional Random Field (CRF) to localize actions. In the above method and typical approaches to this problem, localization is performed in an offline manner where all the video frames are processed together. This prevents timely localization and prediction of actions/interactions - an important consideration for many tasks including surveillance and human-machine interaction. Therefore, in the second part of this dissertation we propose an online approach to the challenging problem of localization and prediction of actions/interactions in videos. In this approach, we use human poses and superpixels in each frame to train discriminative appearance models and perform online prediction of actions/interactions with Structural SVM. Above two approaches rely on human supervision in the form of assigning action class labels to videos and annotating actor bounding boxes in each frame of training videos. Therefore, in the third part of this dissertation we address the problem of unsupervised action localization. Given unlabeled videos without annotations, this approach aims at: 1) Discovering action classes using a discriminative clustering approach, and 2) Localizing actions using a variant of Knapsack problem.
Show less - Date Issued
- 2017
- Identifier
- CFE0006917, ucf:51685
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0006917
- Title
- A MACHINE LEARNING APPROACH TO ASSESS THE SEPARATION OF SEISMOCARDIOGRAPHIC SIGNALS BY RESPIRATION.
- Creator
-
Solar, Brian, Mansy, Hansen, University of Central Florida
- Abstract / Description
-
The clinical usage of Seismocardiography (SCG) is increasing as it is being shown to be an effective non-invasive measurement for heart monitoring. SCG measures the vibrational activity at the chest surface and applications include non-invasive assessment of myocardial contractility and systolic time intervals. Respiratory activity can also affect the SCG signal by changing the hemodynamic characteristics of cardiac activity and displacing the position of the heart. Other clinically...
Show moreThe clinical usage of Seismocardiography (SCG) is increasing as it is being shown to be an effective non-invasive measurement for heart monitoring. SCG measures the vibrational activity at the chest surface and applications include non-invasive assessment of myocardial contractility and systolic time intervals. Respiratory activity can also affect the SCG signal by changing the hemodynamic characteristics of cardiac activity and displacing the position of the heart. Other clinically significant information, such as systolic time intervals, can thus manifest themselves differently in an SCG signal during inspiration and expiration. Grouping SCG signals into their respective respiratory cycle can mitigate this issue. Prior research has focused on developing machine learning classification methods to classify SCG events as according to their respiration cycle. However, recent research at the Biomedical Acoustics Research Laboratory (BARL) at UCF suggests grouping SCG signals into high and low lung volume may be more effective. This research aimed at com- paring the efficiency of grouping SCG signals according to their respiration and lung volume phase and also developing a method to automatically identify the respiration and lung volume phase of SCG events.
Show less - Date Issued
- 2018
- Identifier
- CFH2000310, ucf:45877
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFH2000310
- Title
- USING STUDENT MOOD AND TASK PERFORMANCE TO TRAIN CLASSIFIER ALGORITHMS TO SELECT EFFECTIVE COACHING STRATEGIES WITHIN INTELLIGENT TUTORING SYSTEMS (ITS).
- Creator
-
Sottilare, Robert, Proctor, Michael, University of Central Florida
- Abstract / Description
-
The ultimate goal of this research was to improve student performance by adjusting an Intelligent Tutoring System's (ITS) coaching strategy based on the student's mood. As a step toward this goal, this study evaluated the relationships between each student's mood variables (pleasure, arousal, dominance and mood intensity), the coaching strategy selected by the ITS and the student's performance. Outcomes included methods to increase the perception of the intelligent tutor to...
Show moreThe ultimate goal of this research was to improve student performance by adjusting an Intelligent Tutoring System's (ITS) coaching strategy based on the student's mood. As a step toward this goal, this study evaluated the relationships between each student's mood variables (pleasure, arousal, dominance and mood intensity), the coaching strategy selected by the ITS and the student's performance. Outcomes included methods to increase the perception of the intelligent tutor to allow it to adapt coaching strategies (methods of instruction) to the student's affective needs to mitigate barriers to performance (e.g. negative affect) during the one-to-one tutoring process. The study evaluated whether the affective state (specifically mood) of the student moderated the student's interaction with the tutor and influenced performance. This research examined the relationships, interactions and influences of student mood in the selection of ITS coaching strategies to determine which strategies were more effective in terms of student performance given the student's mood, state (recent sleep time, previous knowledge and training, and interest level) and actions (e.g. mouse movement rate). Two coaching strategies were used in this study: Student-Requested Feedback (SRF) and Tutor-Initiated Feedback (TIF). The SRF coaching strategy provided feedback in the form of hints, questions, direction and support only when the student requested help. The TIF coaching strategy provided feedback (hints, questions, direction or support) at key junctures in the learning process when the student either made progress or failed to make progress in a timely fashion. The relationships between the coaching strategies, mood, performance and other variables of interest were considered in light of five hypotheses. At alpha = .05 and beta at least as great as .80, significant effects were limited in predicting performance. Highlighted findings include no significant differences in the mean performance due to coaching strategies, and only small effect sizes in predicting performance making the regression models developed not of practical significance. However, several variables including performance, energy level and mouse movement rates were significant, unobtrusive predictors of mood. Regression algorithms were developed using Arbuckle's (2008) Analysis of MOment Structures (AMOS) tool to compare the predicted performance for each strategy and then to choose the optimal strategy. A set of production rules were also developed to train a machine learning classifier using Witten & Frank's (2005) Waikato Environment for Knowledge Analysis (WEKA) toolset. The classifier was tested to determine its ability to recognize critical relationships and adjust coaching strategies to improve performance. This study found that the ability of the intelligent tutor to recognize key affective relationships contributes to improved performance. Study assumptions include a normal distribution of student mood variables, student state variables and student action variables and the equal mean performance of the two coaching strategy groups (student-requested feedback and tutor-initiated feedback ). These assumptions were substantiated in the study. Potential applications of this research are broad since its approach is application independent and could be used within ill-defined or very complex domains where judgment might be influenced by affect (e.g. study of the law, decisions involving risk of injury or death, negotiations or investment decisions). Recommendations for future research include evaluation of the temporal, as well as numerical, relationships of student mood, performance, actions and state variables.
Show less - Date Issued
- 2009
- Identifier
- CFE0002528, ucf:47644
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0002528
- Title
- MULTIZOOM ACTIVITY RECOGNITION USING MACHINE LEARNING.
- Creator
-
Smith, Raymond, Shah, Mubarak, University of Central Florida
- Abstract / Description
-
In this thesis we present a system for detection of events in video. First a multiview approach to automatically detect and track heads and hands in a scene is described. Then, by making use of epipolar, spatial, trajectory, and appearance constraints, objects are labeled consistently across cameras (zooms). Finally, we demonstrate a new machine learning paradigm, TemporalBoost, that can recognize events in video. One aspect of any machine learning algorithm is in the feature set used. The...
Show moreIn this thesis we present a system for detection of events in video. First a multiview approach to automatically detect and track heads and hands in a scene is described. Then, by making use of epipolar, spatial, trajectory, and appearance constraints, objects are labeled consistently across cameras (zooms). Finally, we demonstrate a new machine learning paradigm, TemporalBoost, that can recognize events in video. One aspect of any machine learning algorithm is in the feature set used. The approach taken here is to build a large set of activity features, though TemporalBoost itself is able to work with any feature set other boosting algorithms use. We also show how multiple levels of zoom can cooperate to solve problems related to activity recognition.
Show less - Date Issued
- 2005
- Identifier
- CFE0000865, ucf:46658
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0000865
- Title
- AN ADAPTIVE MULTIOBJECTIVE EVOLUTIONARY APPROACH TO OPTIMIZE ARTMAP NEURAL NETWORKS.
- Creator
-
Kaylani, Assem, Georgiopoulos, Michael, University of Central Florida
- Abstract / Description
-
This dissertation deals with the evolutionary optimization of ART neural network architectures. ART (adaptive resonance theory) was introduced by a Grossberg in 1976. In the last 20 years (1987-2007) a number of ART neural network architectures were introduced into the literature (Fuzzy ARTMAP (1992), Gaussian ARTMAP (1996 and 1997) and Ellipsoidal ARTMAP (2001)). In this dissertation, we focus on the evolutionary optimization of ART neural network architectures with the intent of optimizing...
Show moreThis dissertation deals with the evolutionary optimization of ART neural network architectures. ART (adaptive resonance theory) was introduced by a Grossberg in 1976. In the last 20 years (1987-2007) a number of ART neural network architectures were introduced into the literature (Fuzzy ARTMAP (1992), Gaussian ARTMAP (1996 and 1997) and Ellipsoidal ARTMAP (2001)). In this dissertation, we focus on the evolutionary optimization of ART neural network architectures with the intent of optimizing the size and the generalization performance of the ART neural network. A number of researchers have focused on the evolutionary optimization of neural networks, but no research has been performed on the evolutionary optimization of ART neural networks, prior to 2006, when Daraiseh has used evolutionary techniques for the optimization of ART structures. This dissertation extends in many ways and expands in different directions the evolution of ART architectures, such as: (a) uses a multi-objective optimization of ART structures, thus providing to the user multiple solutions (ART networks) with varying degrees of merit, instead of a single solution (b) uses GA parameters that are adaptively determined throughout the ART evolution, (c) identifies a proper size of the validation set used to calculate the fitness function needed for ART's evolution, thus speeding up the evolutionary process, (d) produces experimental results that demonstrate the evolved ART's effectiveness (good accuracy and small size) and efficiency (speed) compared with other competitive ART structures, as well as other classifiers (CART (Classification and Regression Trees) and SVM (Support Vector Machines)). The overall methodology to evolve ART using a multi-objective approach, the chromosome representation of an ART neural network, the genetic operators used in ART's evolution, and the automatic adaptation of some of the GA parameters in ART's evolution could also be applied in the evolution of other exemplar based neural network classifiers such as the probabilistic neural network and the radial basis function neural network.
Show less - Date Issued
- 2008
- Identifier
- CFE0002212, ucf:47907
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0002212
- Title
- Characterization, Classification, and Genesis of Seismocardiographic Signals.
- Creator
-
Taebi, Amirtaha, Mansy, Hansen, Kassab, Alain, Huang, Helen, Vosoughi, Azadeh, University of Central Florida
- Abstract / Description
-
Seismocardiographic (SCG) signals are the acoustic and vibration induced by cardiac activity measured non-invasively at the chest surface. These signals may offer a method for diagnosing and monitoring heart function. Successful classification of SCG signals in health and disease depends on accurate signal characterization and feature extraction.In this study, SCG signal features were extracted in the time, frequency, and time-frequency domains. Different methods for estimating time-frequency...
Show moreSeismocardiographic (SCG) signals are the acoustic and vibration induced by cardiac activity measured non-invasively at the chest surface. These signals may offer a method for diagnosing and monitoring heart function. Successful classification of SCG signals in health and disease depends on accurate signal characterization and feature extraction.In this study, SCG signal features were extracted in the time, frequency, and time-frequency domains. Different methods for estimating time-frequency features of SCG were investigated. Results suggested that the polynomial chirplet transform outperformed wavelet and short time Fourier transforms.Many factors may contribute to increasing intrasubject SCG variability including subject posture and respiratory phase. In this study, the effect of respiration on SCG signal variability was investigated. Results suggested that SCG waveforms can vary with lung volume, respiratory flow direction, or a combination of these criteria. SCG events were classified into groups belonging to these different respiration phases using classifiers, including artificial neural networks, support vector machines, and random forest. Categorizing SCG events into different groups containing similar events allows more accurate estimation of SCG features.SCG feature points were also identified from simultaneous measurements of SCG and other well-known physiologic signals including electrocardiography, phonocardiography, and echocardiography. Future work may use this information to get more insights into the genesis of SCG.
Show less - Date Issued
- 2018
- Identifier
- CFE0007106, ucf:51944
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0007106
- Title
- Effective Task Transfer Through Indirect Encoding.
- Creator
-
Verbancsics, Phillip, Stanley, Kenneth, Sukthankar, Gita, Georgiopoulos, Michael, Garibay, Ivan, University of Central Florida
- Abstract / Description
-
An important goal for machine learning is to transfer knowledge between tasks. For example, learning to play RoboCup Keepaway should contribute to learning the full game of RoboCup soccer. Often approaches to task transfer focus on transforming the original representation to fit the new task. Such representational transformations are necessary because the target task often requires new state information that was not included in the original representation. In RoboCup Keepaway, changing from...
Show moreAn important goal for machine learning is to transfer knowledge between tasks. For example, learning to play RoboCup Keepaway should contribute to learning the full game of RoboCup soccer. Often approaches to task transfer focus on transforming the original representation to fit the new task. Such representational transformations are necessary because the target task often requires new state information that was not included in the original representation. In RoboCup Keepaway, changing from the 3 vs. 2 variant of the task to 4 vs. 3 adds state information for each of the new players. In contrast, this dissertation explores the idea that transfer is most effective if the representation is designed to be the same even across different tasks. To this end, (1) the bird's eye view (BEV) representation is introduced, which can represent different tasks on the same two-dimensional map. Because the BEV represents state information associated with positions instead of objects, it can be scaled to more objects without manipulation. In this way, both the 3 vs. 2 and 4 vs. 3 Keepaway tasks can be represented on the same BEV, which is (2) demonstrated in this dissertation.Yet a challenge for such representation is that a raw two-dimensional map is high-dimensional and unstructured. This dissertation demonstrates how this problem is addressed naturally by the Hypercube-based NeuroEvolution of Augmenting Topologies (HyperNEAT) approach. HyperNEAT evolves an indirect encoding, which compresses the representation by exploiting its geometry. The dissertation then explores further exploiting the power of such encoding, beginning by (3) enhancing the configuration of the BEV with a focus on modularity. The need for further nonlinearity is then (4) investigated through the addition of hidden nodes. Furthermore, (5) the size of the BEV can be manipulated because it is indirectly encoded. Thus the resolution of the BEV, which is dictated by its size, is increased in precision and culminates in a HyperNEAT extension that is expressed at effectively infinite resolution. Additionally, scaling to higher resolutions through gradually increasing the size of the BEV is explored. Finally, (6) the ambitious problem of scaling from the Keepaway task to the Half-field Offense task is investigated with the BEV. Overall, this dissertation demonstrates that advanced representations in conjunction with indirect encoding can contribute to scaling learning techniques to more challenging tasks, such as the Half-field Offense RoboCup soccer domain.
Show less - Date Issued
- 2011
- Identifier
- CFE0004174, ucf:49071
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0004174
- Title
- Data Representation in Machine Learning Methods with its Application to Compilation Optimization and Epitope Prediction.
- Creator
-
Sher, Yevgeniy, Zhang, Shaojie, Dechev, Damian, Leavens, Gary, Gonzalez, Avelino, Zhi, Degui, University of Central Florida
- Abstract / Description
-
In this dissertation we explore the application of machine learning algorithms to compilation phase order optimization, and epitope prediction. The common thread running through these two disparate domains is the type of data being dealt with. In both problem domains we are dealing with categorical data, with its representation playing a significant role in the performance of classification algorithms.We first present a neuroevolutionary approach which orders optimization phases to generate...
Show moreIn this dissertation we explore the application of machine learning algorithms to compilation phase order optimization, and epitope prediction. The common thread running through these two disparate domains is the type of data being dealt with. In both problem domains we are dealing with categorical data, with its representation playing a significant role in the performance of classification algorithms.We first present a neuroevolutionary approach which orders optimization phases to generate compiled programs with performance superior to those compiled using LLVM's -O3 optimization level. Performance improvements calculated as the speed of the compiled program's execution ranged from 27% for the ccbench program, to 40.8% for bzip2.This dissertation then explores the problem of data representation of 3D biological data, such as amino acids. A new approach for distributed representation of 3D biological data through the process of embedding is proposed and explored. Analogously to word embedding, we developed a system that uses atomic and residue coordinates to generate distributed representation for residues, which we call 3D Residue BioVectors. Preliminary results are presented which demonstrate that even the low dimensional 3D Residue BioVectors can be used to predict conformational epitopes and protein-protein interactions, with promising proficiency. The generation of such 3D BioVectors, and the proposed methodology, opens the door for substantial future improvements, and application domains.The dissertation then explores the problem domain of linear B-Cell epitope prediction. This problem domain deals with predicting epitopes based strictly on the protein sequence. We present the DRREP system, which demonstrates how an ensemble of shallow neural networks can be combined with string kernels and analytical learning algorithm to produce state of the art epitope prediction results. DRREP was tested on the SARS subsequence, the HIV, Pellequer, AntiJen datasets, and the standard SEQ194 test dataset. AUC improvements achieved over the state of the art ranged from 3% to 8%.Finally, we present the SEEP epitope classifier, which is a multi-resolution SMV ensemble based classifier which uses conjoint triad feature representation, and produces state of the art classification results. SEEP leverages the domain specific knowledge based protein sequence encoding developed within the protein-protein interaction research domain. Using an ensemble of multi-resolution SVMs, and a sliding window based pre and post processing pipeline, SEEP achieves an AUC of 91.2 on the standard SEQ194 test dataset, a 24% improvement over the state of the art.
Show less - Date Issued
- 2017
- Identifier
- CFE0006793, ucf:51829
- Format
- Document (PDF)
- PURL
- http://purl.flvc.org/ucf/fd/CFE0006793
