Matching Items (143)
Filtering by
- Genre: Academic theses
- Creators: Davulcu, Hasan
Description
Deep neural networks have been shown to be vulnerable to adversarial attacks. Typical attack strategies alter authentic data subtly so as to obtain adversarial samples that resemble the original but otherwise would cause a network's misbehavior such as a high misclassification rate. Various attack approaches have been reported, with some showing state-of-the-art performance in attacking certain networks. In the meanwhile, many defense mechanisms have been proposed in the literature, some of which are quite effective for guarding against typical attacks. Yet, most of these attacks fail when the targeted network modifies its architecture or uses another set of parameters and vice versa. Moreover, the emerging of more advanced deep neural networks, such as generative adversarial networks (GANs), has made the situation more complicated and the game between the attack and defense is continuing. This dissertation aims at exploring the venerability of the deep neural networks by investigating the mechanisms behind the success/failure of the existing attack and defense approaches. Therefore, several deep learning-based approaches have been proposed to study the problem from different perspectives. First, I developed an adversarial attack approach by exploring the unlearned region of a typical deep neural network which is often over-parameterized. Second, I proposed an end-to-end learning framework to analyze the images generated by different GAN models. Third, I developed a defense mechanism that can secure the deep neural network against adversarial attacks with a defense layer consisting of a set of orthogonal kernels. Substantial experiments are conducted to unveil the potential factors that contribute to attack/defense effectiveness. This dissertation also concludes with a discussion of possible future works of achieving a robust deep neural network.
ContributorsDing, Yuzhen (Author) / Li, Baoxin (Thesis advisor) / Davulcu, Hasan (Committee member) / Venkateswara, Hemanth Kumar Demakethepalli (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2022
Description
Bitcoin (BTC) shares many characteristics with traditional stocks, but it is much more volatile since the cryptocurrency market is unregulated. The high volatility makes BTC a very high risk-high reward investment and predictive analysis can be very useful to obtain good returns and minimize risk. Taking Cocco et al. [1] as the primary reference, this thesis tries to reproduce their findings by building two BTC price forecasting models, Long Short-Term Memory (LSTM) and Bayesian Neural Network (BNN), and finding that the Mean Absolute Percentage Error (MAPE) is lower for the initial BNN model in comparison to the initial LSTM model. In addition to forecasting the value of BTC, a metric called trend% is developed to gauge the models’ ability to capture the trend of how the price varies from one timestep to the next and used to compare the trend prediction performance. It is found that both initial models make random predictions for the trend. Improvements like removing the stochastic component from the data and forecasting returns as opposed to price values show that both models show comparable performance in terms of both MAPE and trend%. The thesis concludes by discussing the future work that can be done to potentially improve the above models. One of the possibilities mentioned is to use on-chain data from the BTC blockchain coupled with the real-world knowledge of BTC exchanges and feed this as input features to the models.
ContributorsMittal, Shivansh (Author) / Boscovic, Dragan (Thesis advisor) / Davulcu, Hasan (Committee member) / Candan, Kasim (Committee member) / Arizona State University (Publisher)
Created2022
Description
In some scenarios, true temporal ordering is required to identify the actions occurring in a video. Recently a new synthetic dataset named CATER, was introduced containing 3D objects like sphere, cone, cylinder etc. which undergo simple movements such as slide, pick & place etc. The task defined in the dataset is to identify compositional actions with temporal ordering. In this thesis, a rule-based system and a window-based technique are proposed to identify individual actions (atomic) and multiple actions with temporal ordering (composite) on the CATER dataset. The rule-based system proposed here is a heuristic algorithm that evaluates the magnitude and direction of object movement across frames to determine the atomic action temporal windows and uses these windows to predict the composite actions in the videos. The performance of the rule-based system is validated using the frame-level object coordinates provided in the dataset and it outperforms the performance of the baseline models on the CATER dataset. A window-based training technique is proposed for identifying composite actions in the videos. A pre-trained deep neural network (I3D model) is used as a base network for action recognition. During inference, non-overlapping windows are passed through the I3D network to obtain the atomic action predictions and the predictions are passed through a rule-based system to determine the composite actions. The approach outperforms the state-of-the-art composite action recognition models by 13.37% (mAP 66.47% vs. mAP 53.1%).
ContributorsMaskara, Vivek Kumar (Author) / Venkateswara, Hemanth (Thesis advisor) / McDaniel, Troy (Thesis advisor) / Davulcu, Hasan (Committee member) / Arizona State University (Publisher)
Created2022
Description
Large Language Models (LLMs) have displayed impressive capabilities in handling tasks that require few demonstration examples, making them effective few-shot learn- ers. Despite their potential, LLMs face challenges when it comes to addressing com- plex real-world tasks that involve multiple modalities or reasoning steps. For example, predicting cancer patients’ survival period based on clinical data, cell slides, and ge- nomics poses significant logistical complexities. Although several approaches have been proposed to tackle these challenges, they often fall short in achieving promising performance due to their inability to consider all modalities simultaneously or account for missing modalities, variations in modalities, and the integration of multi-modal data, ultimately compromising their effectiveness.This thesis proposes a novel approach for multi-modal tumor survival prediction to address these limitations. Taking inspiration from recent advancements in LLMs, particularly Mixture of Experts (MoE)-based models, a graph-guided MoE framework is introduced. This framework utilizes a graph structure to manage the predictions effectively and combines multiple models to enhance predictive power. Rather than training a single foundation model for end-to-end survival prediction, the approach leverages a MOE-guided ensemble to manage model callings as tools automatically. By leveraging the strengths of existing models and guiding them through a MOE framework, the aim is to achieve better performance and more accurate predictions in complex real-world tasks. Experiments and analysis on the TCGA-LUAD dataset show improved performance over the individual modal and vanilla ensemble models.
ContributorsMathavan, Hirthik (Author) / Liu, Huan (Thesis advisor) / Davulcu, Hasan (Committee member) / Choi, YooJung (Committee member) / Arizona State University (Publisher)
Created2024
Description
Animal pose estimation (APE) is utilized in preclinical research settings for various neurological disorders such as Parkinson's disease (PD), Huntington's disease (HD) and multiple sclerosis. The technique includes real-time scoring of impairment in the animals during testing or video recording. This is a time-consuming operation prone to errors due to visual fatigue. To overcome these shortcomings, APE automation by deep learning has been studied. The field of APE has gone through significant development backed by improvements in deep learning techniques. These developments have improved 2D and 3D pose estimation, 3D mesh reconstruction and behavior prediction capabilities. As a result, there are numerous sophisticated tools and datasets available today. Despite these developments, APE still lags behind human observer scoring with respect to accuracy and flexibility under complex scenarios. In this project, two critical challenges are being addressed within the context of neurological research focusing on PD. The first challenge is about the lack of comprehensive diverse datasets necessary for accurate training as well as for fine-tuning deep learning models. This is compounded by the inherent difficulty in working with uncooperative rodent subjects, whose unpredictable behaviors often impede reliable data collection. The second challenge focuses on reduction in variation of scores that result from being scored by different evaluators. This will also involve tackling bias and reducing human error for the purpose of reliable and accurate assessments. In order to address these issues, systematic data collection and deep learning in APE have been utilized to automate manual scoring procedures. This project will contribute to neurological research, particularly in understanding and treating disorders like PD. The goal is to improve methods used in assessing rodent behavior which could aid in developing effective therapeutics. The successful implementation of an automated scoring mechanism could set a new standard in neurological research, offering insights and methodologies that are more accurate and reliable.
ContributorsJanapareddi, Ajay Kumar (Author) / Davulcu, Hasan (Thesis advisor) / Poste, George (Thesis advisor) / Kelley, Christy (Committee member) / Arizona State University (Publisher)
Created2024
Description
With the exponential growth of multi-modal data in the field of computer vision, the ability to do inference effectively among multiple modalities—such as visual, textual, and auditory data—shows significant opportunities. The rapid development of cross-modal applications such as retrieval and association is primarily attributed to their ability to bridge the gap between different modalities of data. However, the current mainstream cross-modal methods always heavily rely on the availability of fully annotated paired data, presenting a significant challenge due to the scarcity of precisely matched datasets in real-world scenarios. In response to this bottleneck, several sophisticated deep learning algorithms are designed to substantially improve the inference capabilities across a broad spectrum of cross-modal applications. This dissertation introduces novel deep learning algorithms aimed at enhancing inference capabilities in cross-modal applications, which take four primary aspects. Firstly, it introduces the algorithm for image retrieval by learning hashing codes. This algorithm only utilizes the other modality data in weakly supervised tags format rather than the supervised label. Secondly, it designs a novel framework for learning the joint embeddings of images and texts for the cross-modal retrieval tasks. It efficiently learns the binary codes from the continuous CLIP feature space and can even deliver competitive performance compared with the results from non-hashing methods. Thirdly, it conducts a method to learn the fragment-level embeddings that capture fine-grained cross-modal association in images and texts. This method uses the fragment proposals in an unsupervised manner. Lastly, this dissertation also outlines the algorithm to enhance the mask-text association ability of pre-trained semantic segmentation models with zero examples provided. Extensive future plans to further improve this algorithm for semantic segmentation tasks will be discussed.
ContributorsZhuo, Yaoxin (Author) / Li, Baoxin (Thesis advisor) / Wu, Teresa (Committee member) / Davulcu, Hasan (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2024
Description
Experience, whether personal or vicarious, plays an influential role in shaping human knowledge. Through these experiences, one develops an understanding of the world, which leads to learning. The process of gaining knowledge in higher education transcends beyond the passive transmission of knowledge from an expert to a novice. Instead, students are encouraged to actively engage in every learning opportunity to achieve mastery in their chosen field. Evaluation of such mastery typically entails using educational assessments that provide objective measures to determine whether the student has mastered what is required of them. With the proliferation of educational technology in the modern classroom, information about students is being collected at an unprecedented rate, covering demographic, performance, and behavioral data. In the absence of analytics expertise, stakeholders may miss out on valuable insights that can guide future instructional interventions, especially in helping students understand their strengths and weaknesses. This dissertation presents Web-Programming Grading Assistant (WebPGA), a homegrown educational technology designed based on various learning sciences principles, which has been used by 6,000+ students. In addition to streamlining and improving the grading process, it encourages students to reflect on their performance. WebPGA integrates learning analytics into educational assessments using students' physical and digital footprints. A series of classroom studies is presented demonstrating the use of learning analytics and assessment data to make students aware of their misconceptions. It aims to develop ways for students to learn from previous mistakes made by themselves or by others. The key findings of this dissertation include the identification of effective strategies of better-performing students, the demonstration of the importance of individualized guidance during the reviewing process, and the likely impact of validating one's understanding of another's experiences. Moreover, the Personalized Recommender of Items to Master and Evaluate (PRIME) framework is introduced. It is a novel and intelligent approach for diagnosing one's domain mastery and providing tailored learning opportunities by allowing students to observe others' mistakes. Thus, this dissertation lays the groundwork for further improvement and inspires better use of available data to improve the quality of educational assessments that will benefit both students and teachers.
ContributorsParedes, Yancy Vance (Author) / Hsiao, I-Han (Thesis advisor) / VanLehn, Kurt (Thesis advisor) / Craig, Scotty D (Committee member) / Bansal, Srividya (Committee member) / Davulcu, Hasan (Committee member) / Arizona State University (Publisher)
Created2023
Description
Modern data center networks require efficient and scalable security analysis approaches that can analyze the relationship between the vulnerabilities. Utilizing the Attack Representation Methods (ARMs) and Attack Graphs (AGs) enables the security administrator to understand the cloud network’s current security situation at the low-level. However, the AG approach suffers from scalability challenges. It relies on the connectivity between the services and the vulnerabilities associated with the services to allow the system administrator to realize its security state. In addition, the security policies created by the administrator can have conflicts among them, which is often detected in the data plane of the Software Defined Networking (SDN) system. Such conflicts can cause security breaches and increase the flow rules processing delay. This dissertation addresses these challenges with novel solutions to tackle the scalability issue of Attack Graphs and detect security policy conflictsin the application plane before they are transmitted into the data plane for final
installation. Specifically, it introduces a segmentation-based scalable security state
(S3) framework for the cloud network. This framework utilizes the well-known divide-and-conquer approach to divide the large network region into smaller, manageable segments. It follows a well-known segmentation approach derived from the K-means clustering algorithm to partition the system into segments based on the similarity between the services. Furthermore, the dissertation presents unified intent rules that abstract the network administration from the underlying network controller’s format. It develops a networking service solution to use a bounded formal model for network service compliance checking that significantly reduces the complexity of flow rule conflict checking at the data plane level. The solution can be expended from a single SDN domain to multiple SDN domains and hybrid networks by applying network service function chaining (SFC) for inter-domain policy management.
ContributorsSabur, Abdulhakim (Author) / Zhao, Ming (Thesis advisor) / Xue, Guoliang (Committee member) / Davulcu, Hasan (Committee member) / Zhang, Yanchao (Committee member) / Arizona State University (Publisher)
Created2023
Description
Social media platforms have become widely used for open communication, yet their lack of moderation has led to the proliferation of harmful content, including hate speech. Manual monitoring of such vast amounts of user-generated data is impractical, thus necessitating automated hate speech detection methods. Pre-trained language models have been proven to possess strong base capabilities, which not only excel at in-distribution language modeling but also show powerful abilities in out-of-distribution language modeling, transfer learning and few-shot learning. However, these models operate as complex function approximators, mapping input text to a hate speech classification, without providing any insights into the reasoning behind their predictions. Hence, existing methods often lack transparency, hindering their effectiveness, particularly in sensitive content moderation contexts. Recent efforts have been made to integrate their capabilities with large language models like ChatGPT and Llama2, which exhibit reasoning capabilities and broad knowledge utilization. This thesis explores leveraging the reasoning abilities of large language models to enhance the interpretability of hate speech detection. A novel framework is proposed that utilizes state-of-the-art Large Language Models (LLMs) to extract interpretable rationales from input text, highlighting key phrases or sentences relevant to hate speech classification. By incorporating these rationale features into a hate speech classifier, the framework inherently provides transparent and interpretable results. This approach combines the language understanding prowess of LLMs with the discriminative power of advanced hate speech classifiers, offering a promising solution to the challenge of interpreting automated hate speech detection models.
ContributorsNirmal, Ayushi (Author) / Liu, Huan (Thesis advisor) / Davulcu, Hasan (Committee member) / Wei, Hua (Committee member) / Arizona State University (Publisher)
Created2024
Description
Multivariate timeseries data are highly common in the healthcare domain, especially in the neuroscience field for detecting and predicting seizures to monitoring intracranial hypertension (ICH). Unfortunately, conventional techniques to leverage the available time series data do not provide high degrees of accuracy. To address this challenge, the dissertation focuses on onset prediction models for children with brain trauma in collaboration with neurologists at Phoenix Children’s Hospital. The dissertation builds on the key hypothesis that leveraging spatial information underlying the electroencephalogram (EEG) sensor graphs can significantly boost the accuracy in a multi-modal environment, integrating EEG with intracranial pressure (ICP), arterial blood pressure (ABP) and electrocardiogram (ECG) modalities. Based on this key hypothesis, the dissertation focuses on novel metadata supported multi-variate time series analysis algorithms for onset detection and prediction. In particular, the dissertation investigates a model architecture with a dual attention mechanism to draw global dependencies between inputs and outputs, leveraging self-attention in EEG data using multi-head attention for transformers, and long short-term memory (LSTM). However, recognizing that the positional encoding used traditionally in transformers does not help capture the spatial/neighborhood context of EEG sensors, the dissertation investigates novel attention techniques for performing explicit spatial learning using a coupled model network. This dissertation has answered the question of leveraging transformers and LSTM to perform implicit and explicit learning using a metadata supported coupled model network a) Robust Multi-variate Temporal Features (RMT) model and LSTM, b) the convolutional neural network - scale space attention (CNN-SSA) and LSTM mapped together using Multi-Head Attention with explicit spatial metadata for EEG sensor graphs for seizure and ICH onset prediction respectively. In addition, this dissertation focuses on transfer learning between multiple groups where target patients have lesser number of EEG channels than the source patients. This incomplete data poses problems during pre-processing. Two approaches are explored using all predictors approach considering spatial context to guide the variates who are used as predictors for the missing EEG channels, and common core/subset of EEG channels. Under data imputation K-Nearest Neighbors (KNN) regression and multi-variate multi-scale neural network (M2NN) are implemented, to address the problem for target patients.
ContributorsRavindranath, Manjusha (Author) / Candan, K. Selcuk (Thesis advisor) / Davulcu, Hasan (Committee member) / Zou, Jia (Committee member) / Luisa Sapino, Maria (Committee member) / Arizona State University (Publisher)
Created2024