Computer Sciences Commons^™

Action Recognition Model Of Directed Attention Based On Cosine Similarity, Chen Li, Ming He, Chen Dong, Wei Li

Journal of System Simulation

Abstract: Aiming at the lack of directionality of traditional dot product attention, this paper proposes a directed attention model (DAM) based on cosine similarity. To effectively represent the direction relationship between the spatial and temporal features of video frames, the paper defines the relationship function in the attention mechanism using the cosine similarity theory, which can remove the absolute value of the relationship between features. To reduce the computational burden of the attention mechanism, the operation is decomposed from two dimensions of time and space. The computational complexity is further optimized by combining linear attention operation. The experiment is divided …

Machine Learning As A Tool For Early Detection: A Focus On Late-Stage Colorectal Cancer Across Socioeconomic Spectrums, Hadiza Galadima, Rexford Anson-Dwamena, Ashley Johnson, Ghalib Bello, Georges Adunlin, James Blando

Community & Environmental Health Faculty Publications

Purpose: To assess the efficacy of various machine learning (ML) algorithms in predicting late-stage colorectal cancer (CRC) diagnoses against the backdrop of socio-economic and regional healthcare disparities. Methods: An innovative theoretical framework was developed to integrate individual- and census tract-level social determinants of health (SDOH) with sociodemographic factors. A comparative analysis of the ML models was conducted using key performance metrics such as AUC-ROC to evaluate their predictive accuracy. Spatio-temporal analysis was used to identify disparities in late-stage CRC diagnosis probabilities. Results: Gradient boosting emerged as the superior model, with the top predictors for late-stage CRC diagnosis being anatomic site, …

Malware Detection With Artificial Intelligence: A Systematic Literature Review, Matthew G. Gaber, Mohiuddin Ahmed, Helge Janicke

Research outputs 2022 to 2026

In this survey, we review the key developments in the field of malware detection using AI and analyze core challenges. We systematically survey state-of-the-art methods across five critical aspects of building an accurate and robust AI-powered malware-detection model: malware sophistication, analysis techniques, malware repositories, feature selection, and machine learning vs. deep learning. The effectiveness of an AI model is dependent on the quality of the features it is trained with. In turn, the quality and authenticity of these features is dependent on the quality of the dataset and the suitability of the analysis tool. Static analysis is fast but is …

Nonuniform Sampling-Based Breast Cancer Classification, Santiago Posso

Theses and Dissertations--Electrical and Computer Engineering

The emergence of deep learning models and their success in visual object recognition have fueled the medical imaging community's interest in integrating these algorithms to improve medical diagnosis. However, natural images, which have been the main focus of deep learning models and mammograms, exhibit fundamental differences. First, breast tissue abnormalities are often smaller than salient objects in natural images. Second, breast images have significantly higher resolutions but are generally heavily downsampled to fit these images to deep learning models. Models that handle high-resolution mammograms require many exams and complex architectures. Additionally, spatially resizing mammograms leads to losing discriminative details essential …

A Survey On Few-Shot Class-Incremental Learning, Songsong Tian, Lusi Li, Weijun Li, Hang Ran, Xin Ning, Prayag Tiwari

Computer Science Faculty Publications

Large deep learning models are impressive, but they struggle when real-time data is not available. Few-shot class-incremental learning (FSCIL) poses a significant challenge for deep neural networks to learn new tasks from just a few labeled samples without forgetting the previously learned ones. This setup can easily leads to catastrophic forgetting and overfitting problems, severely affecting model performance. Studying FSCIL helps overcome deep learning model limitations on data volume and acquisition time, while improving practicality and adaptability of machine learning models. This paper provides a comprehensive survey on FSCIL. Unlike previous surveys, we aim to synthesize few-shot learning and incremental …

Urban Flood Extent Segmentation And Evaluation From Real-World Surveillance Camera Images Using Deep Convolutional Neural Network, Yidi Wang, Yawen Shen, Behrouz Salahshour, Mecit Cetin, Khan Iftekharuddin, Navid Tahvildari, Guoping Huang, Devin K. Harris, Kwame Ampofo, Jonathan L. Goodall

Civil & Environmental Engineering Faculty Publications

This study explores the use of Deep Convolutional Neural Network (DCNN) for semantic segmentation of flood images. Imagery datasets of urban flooding were used to train two DCNN-based models, and camera images were used to test the application of the models with real-world data. Validation results show that both models extracted flood extent with a mean F1-score over 0.9. The factors that affected the performance included still water surface with specular reflection, wet road surface, and low illumination. In testing, reduced visibility during a storm and raindrops on surveillance cameras were major problems that affected the segmentation of flood extent. …

Deep Learning Uncertainty Quantification For Clinical Text Classification, Alina Peluso, Ioana Danciu, Hong-Jun Yoon, Jamaludin Mohd Yusof, Tanmoy Bhattacharya, Adam Spannaus, Noah Schaefferkoetter, Eric B. Durbin, Xiao-Cheng Wu, Antoinette Stroup, Jennifer Doherty, Stephen Schwartz, Charles Wiggins, Linda Coyle, Lynne Penberthy, Georgia D. Tourassi, Shang Gao

School of Public Health Faculty Publications

INTRODUCTION: Machine learning algorithms are expected to work side-by-side with humans in decision-making pipelines. Thus, the ability of classifiers to make reliable decisions is of paramount importance. Deep neural networks (DNNs) represent the state-of-the-art models to address real-world classification. Although the strength of activation in DNNs is often correlated with the network's confidence, in-depth analyses are needed to establish whether they are well calibrated. METHOD: In this paper, we demonstrate the use of DNN-based classification tools to benefit cancer registries by automating information extraction of disease at diagnosis and at surgery from electronic text pathology reports from the US National …

Data-Centric Image Super-Resolution In Magnetic Resonance Imaging: Challenges And Opportunities, Mamata Shrestha

Graduate Theses and Dissertations

Super-resolution has emerged as a crucial research topic in the field of Magnetic Resonance Imaging (MRI) where it plays an important role in understanding and analysis of complex, qualitative, and quantitative characteristics of tissues at high resolutions. Deep learning techniques have been successful in achieving state-of-the-art results for super-resolution. These deep learning-based methods heavily rely on a substantial amount of data. Additionally, they require a pair of low-resolution and high-resolution images for supervised training which is often unavailable. Particularly in MRI super-resolution, it is often impossible to have low-resolution and high-resolution training image pairs. To overcome this, existing methods for …

Distxplore: Distribution-Guided Testing For Evaluating And Enhancing Deep Learning Systems, Longtian Wang, Xiaofei Xie, Xiaoning Du, Meng Tian, Qing Guo, Zheng Yang, Chao Shen

Research Collection School Of Computing and Information Systems

Deep learning (DL) models are trained on sampled data, where the distribution of training data differs from that of real-world data (i.e., the distribution shift), which reduces the model's robustness. Various testing techniques have been proposed, including distribution-unaware and distribution-aware methods. However, distribution-unaware testing lacks effectiveness by not explicitly considering the distribution of test cases and may generate redundant errors (within same distribution). Distribution-aware testing techniques primarily focus on generating test cases that follow the training distribution, missing out-of-distribution data that may also be valid and should be considered in the testing process. In this paper, we propose a novel …

A Systematic Collection Of Medical Image Datasets For Deep Learning, Johann Li, Guangming Zhu, Cong Hua, Mingtao Feng, Basheer Bennamoun, Ping Li, Xiaoyuan Lu, Juan Song, Peiyi Shen, Xu Xu, Lin Mei, Liang Zhang, Syed A. A. Shah, Mohammed Bennamoun

Research outputs 2022 to 2026

The astounding success made by artificial intelligence in healthcare and other fields proves that it can achieve human-like performance. However, success always comes with challenges. Deep learning algorithms are data dependent and require large datasets for training. Many junior researchers face a lack of data for a variety of reasons. Medical image acquisition, annotation, and analysis are costly, and their usage is constrained by ethical restrictions. They also require several other resources, such as professional equipment and expertise. That makes it difficult for novice and non-medical researchers to have access to medical data. Thus, as comprehensively as possible, this article …

Uavs And Deep Neural Networks: An Alternative Approach To Monitoring Waterfowl At The Site Level, Zachary J. Loken

LSU Master's Theses

Understanding how waterfowl respond to habitat restoration and management activities is crucial for evaluating and refining conservation delivery programs. However, site-specific waterfowl monitoring is challenging, especially in heavily forested systems such as the Mississippi Alluvial Valley (MAV)—a primary wintering region for ducks in North America. I hypothesized that using uncrewed aerial vehicles (UAVs) coupled with deep learning-based methods for object detection would provide an efficient and effective means for surveying non-breeding waterfowl on difficult-to-access restored wetland sites. Accordingly, during the winters of 2021 and 2022, I surveyed wetland restoration easements in the MAV using a UAV equipped with a dual …

Understanding The Impact Of Trade Policy Effect Uncertainty On Firm-Level Innovation Investment: A Deep Learning Approach, Daniel. Chen, Nan Hu, Peng. Liang, Morgan. Swink

Research Collection School Of Computing and Information Systems

Integrating the real options perspective and resource dependence theory, this study examines how firms adjust their innovation investments to trade policy effect uncertainty (TPEU), a less studied type of firm specific, perceived environmental uncertainty in which managers have difficulty predicting how potential policy changes will affect business operations. To develop a text-based, context-dependent, time-varying measure of firm-level perceived TPEU, we apply Bidirectional Encoder Representations from Transformers (BERT), a state-of-the-art deep learning approach. We apply BERT to analyze the texts of mandatory Management Discussion and Analysis (MD&A) sections of annual reports for a sample of 22,669 firm-year observations from 3,181 unique …

Optimized Uncertainty Estimation For Vision Transformers: Enhancing Adversarial Robustness And Performance Using Selective Classification, Erik Pautsch, John Li, Silvio Rizzi, George K. Thiruvathukal, Maria Pantoja

Computer Science: Faculty Publications and Other Works

Deep Learning models often exhibit undue confidence when encountering out-of-distribution (OOD) inputs, misclassifying with high confidence. The ideal outcome, in these cases, would be an "I do not know" verdict. We enhance the trustworthiness of our models through selective classification, allowing the model to abstain from making predictions when facing uncertainty. Rather than a singular prediction, the model offers a prediction distribution, enabling users to gauge the model’s trustworthiness and determine the need for human intervention. We assess uncertainty in two baseline models: a Convolutional Neural Network (CNN) and a Vision Transformer (ViT). By leveraging these uncertainty values, we minimize …

Object Recognition With Deep Neural Networks In Low-End Systems, Lillian Davis

Mahurin Honors College Capstone Experience/Thesis Projects

Object recognition is an important area in computer vision. Object recognition has been advanced significantly by deep learning that unifies feature extraction and classification. In general, deep neural networks, such as Convolution Neural Networks (CNNs), are trained in high-performance systems. Aiming to extend the reach of deep learning to personal computing, I propose a study of deep learning-based object recognition in low-end systems, such as laptops. This research includes how differing layer configurations and hyperparameter values used in CNNs can either create or resolve the issue of overfitting and affect final accuracy levels of object recognition systems. The main contribution …

An Ai-Based Framework For Translating American Sign Language To English And Vice Versa, Vijayendra D. Avina, Md Amiruzzaman, Stefanie Amiruzzaman, Linh B. Ngo, M. Ali Akber Dewan

Computer Science Faculty Publications

Abstract: In this paper, we propose a framework to convert American Sign Language (ASL) to English and English to ASL. Within this framework, we use a deep learning model along with the rolling average prediction that captures image frames from videos and classifies the signs from the image frames. The classified frames are then used to construct ASL words and sentences to support people with hearing impairments. We also use the same deep learning model to capture signs from the people with deaf symptoms and convert them into ASL words and English sentences. Based on this framework, we developed a …

Spoken Language Processing And Modeling For Aviation Communications, Aaron Van De Brook

Doctoral Dissertations and Master's Theses

With recent advances in machine learning and deep learning technologies and the creation of larger aviation-specific corpora, applying natural language processing technologies, especially those based on transformer neural networks, to aviation communications is becoming increasingly feasible. Previous work has focused on machine learning applications to natural language processing, such as N-grams and word lattices. This thesis experiments with a process for pretraining transformer-based language models on aviation English corpora and compare the effectiveness and performance of language models transfer learned from pretrained checkpoints and those trained from their base weight initializations (trained from scratch). The results suggest that transformer language …

Faster, Cheaper, And Better Cfd: A Case For Machine Learning To Augment Reynolds-Averaged Navier-Stokes, John Peter Romano Ii

Mechanical & Aerospace Engineering Theses & Dissertations

In recent years, the field of machine learning (ML) has made significant advances, particularly through applying deep learning (DL) algorithms and artificial intelligence (AI). The literature shows several ways that ML may enhance the power of computational fluid dynamics (CFD) to improve its solution accuracy, reduce the needed computational resources and reduce overall simulation cost. ML techniques have also expanded the understanding of underlying flow physics and improved data capture from experimental fluid dynamics.

This dissertation presents an in-depth literature review and discusses ways the field of fluid dynamics has leveraged ML modeling to date. The author selects and describes …

Style Transfer Network For Generating Opera Makeup Details, Fengquan Zhang, Duo Cao, Xiaohan Ma, Baijun Chen, Jiangxiao Zhang

Journal of System Simulation

Abstract: To address the problem of the loss of local style details in cross-domain image simulation, a ChinOperaGAN network framework suitable for opera makeup is designed from the perspective of protecting the excellent traditional culture. In order to solve the style translation of differences in two image domains, multiple overlapping local adversarial discriminators are proposed in the generative adversarial network. Since paired opera makeup data are difficult to obtain, a synthetic image is generated by combining the source image makeup mapping to effectively guide the transfer of local makeup details between images. In view of the characteristics of opera makeup …

Rosas: Deep Semi-Supervised Anomaly Detection With Contamination-Resilient Continuous Supervision, Hongzuo Xu, Yijie Wang, Guansong Pang, Songlei Jian, Ning Liu, Yongjun Wang

Research Collection School Of Computing and Information Systems

Semi-supervised anomaly detection methods leverage a few anomaly examples to yield drastically improved performance compared to unsupervised models. However, they still suffer from two limitations: 1) unlabeled anomalies (i.e., anomaly contamination) may mislead the learning process when all the unlabeled data are employed as inliers for model training; 2) only discrete supervision information (such as binary or ordinal data labels) is exploited, which leads to suboptimal learning of anomaly scores that essentially take on a continuous distribution. Therefore, this paper proposes a novel semi-supervised anomaly detection method, which devises contamination-resilient continuous supervisory signals. Specifically, we propose a mass interpolation method …

Self-Supervised Pretraining And Transfer Learning On Fmri Data With Transformers, Sean Paulsen

Dartmouth College Ph.D Dissertations

Transfer learning is a machine learning technique founded on the idea that knowledge acquired by a model during “pretraining” on a source task can be transferred to the learning of a target task. Successful transfer learning can result in improved performance, faster convergence, and reduced demand for data. This technique is particularly desirable for the task of brain decoding in the domain of functional magnetic resonance imaging (fMRI), wherein even the most modern machine learning methods can struggle to decode labelled features of brain images. This challenge is due to the highly complex underlying signal, physical and neurological differences between …

Optimizing Collective Communication For Scalable Scientific Computing And Deep Learning, Jiali Li

Doctoral Dissertations

In the realm of distributed computing, collective operations involve coordinated communication and synchronization among multiple processing units, enabling efficient data exchange and collaboration. Scientific applications, such as simulations, computational fluid dynamics, and scalable deep learning, require complex computations that can be parallelized across multiple nodes in a distributed system. These applications often involve data-dependent communication patterns, where collective operations are critical for achieving high performance in data exchange. Optimizing collective operations for scientific applications and deep learning involves improving the algorithms, communication patterns, and data distribution strategies to minimize communication overhead and maximize computational efficiency.

Within the context of this …

The Development Of Artificial Intelligence-Based Tools For Expert Peer Review Of Radiotherapy Treatment Plans, Mary Gronberg

Dissertations & Theses (Open Access)

Creating a patient-specific radiation treatment plan is a time-consuming and operator-dependent manual process. The treatment planner adjusts the planning parameters in a trial-and-error fashion in an effort to balance the competing clinical objectives of tumor coverage and normal tissue sparing. Often, a plan is selected because it meets basic organ at risk dose thresholds for severe toxicity; however, it is evident that a plan with a decreased risk of normal tissue complication probability could be achieved. This discrepancy between “acceptable” and “best possible” plan is magnified if either the physician or treatment planner lacks focal expertise in the disease site. …

Evaluating Chatgpt For Recommendation: How Does The Ability To Converse Impact Recommendation?, Kyle Spurlock

Electronic Theses and Dissertations

Recommendation algorithms have become an absolute necessity in the modern world to avoid information overload. However, the interaction between the human and the system is largely superficial and without any real contact. If you are given poor recommendations, you have no choice but to sift through mountains of content on your own until the model learns to accommodate your tastes more. This is bad for business as well as the consumer. Recently, large language models like ChatGPT have seen a significant rise in popularity due to their ease of use and wide range of knowledge. It has now become nearly …

Multi-Granularity Detector For Vulnerability Fixes, Truong Giang Nguyen, Cong, Thanh Le, Hong Jin Kang, Ratnadira Widyasari, Chengran Yang, Zhipeng Zhao, Bowen Xu, Jiayuan Zhou, Xin Xia, Ahmed E. Hassan, David Lo, David Lo

Research Collection School Of Computing and Information Systems

With the increasing reliance on Open Source Software, users are exposed to third-party library vulnerabilities. Software Composition Analysis (SCA) tools have been created to alert users of such vulnerabilities. SCA requires the identification of vulnerability-fixing commits. Prior works have proposed methods that can automatically identify such vulnerability-fixing commits. However, identifying such commits is highly challenging, as only a very small minority of commits are vulnerability fixing. Moreover, code changes can be noisy and difficult to analyze. We observe that noise can occur at different levels of detail, making it challenging to detect vulnerability fixes accurately. To address these challenges and …

Bertastic At Semeval-2023 Task 3: Fine-Tuning Pretrained Multilingual Transformers – Does Order Matter?, Tarek Mahmoud, Preslav Nakov

Natural Language Processing Faculty Publications

The naïve approach for fine-tuning pretrained deep learning models on downstream tasks involves feeding them mini-batches of randomly sampled data. In this paper, we propose a more elaborate method for fine-tuning Pretrained Multilingual Transformers (PMTs) on multilingual data. Inspired by the success of curriculum learning approaches, we investigate the significance of fine-tuning PMTs on multilingual data in a sequential fashion language by language. Unlike the curriculum learning paradigm where the model is presented with increasingly complex examples, we do not adopt a notion of “easy” and “hard” samples. Instead, our experiments draw insight from psychological findings on how the human …

Learning Deep Time-Index Models For Time Series Forecasting, Jiale Gerald Woo, Chenghao Liu, Doyen Sahoo, Akshat Kumar, Steven Hoi

Research Collection School Of Computing and Information Systems

Deep learning has been actively applied to time series forecasting, leading to a deluge of new methods, belonging to the class of historicalvalue models. Yet, despite the attractive properties of time-index models, such as being able to model the continuous nature of underlying time series dynamics, little attention has been given to them. Indeed, while naive deep timeindex models are far more expressive than the manually predefined function representations of classical time-index models, they are inadequate for forecasting, being unable to generalize to unseen time steps due to the lack of inductive bias. In this paper, we propose DeepTime, a …

Utilizing Few-Shot Meta Learning Algorithms For Medical Image Segmentation, Nick Littlefield

Thinking Matters Symposium

Deep learning models can be difficult to train because they require large amounts of data, which we usually do not have or are too expensive to get or annotate. To overcome this problem, we can use few-shot meta-learning, which allows us to train deep learning models with little data. Using a few examples, meta-learning, or learning-to-learn, aims to use the experience learned during training to generalize to unknown tasks. Medical imaging is an industry where it is particularly useful, as there is limited publicly available data due to patient privacy concerns and annotating costs.

This project examines how meta-learning performs …

Sarcasm Detection In English And Arabic Tweets Using Transformer Models, Rishik Lad

Computer Science Senior Theses

This thesis describes our approach toward the detection of sarcasm and its various types in English and Arabic Tweets through methods in deep learning. There are five problems we attempted: (1) detection of sarcasm in English Tweets, (2) detection of sarcasm in Arabic Tweets, (3) determining the type of sarcastic speech subcategory for English Tweets, (4) determining which of two semantically equivalent English Tweets is sarcastic, and (5) determining which of two semantically equivalent Arabic Tweets is sarcastic. All tasks were framed as classification problems, and our contributions are threefold: (a) we developed an English binary classifier system with RoBERTa, …

Patient Movement Monitoring Based On Imu And Deep Learning, Mohsen Sharifi Renani

Electronic Theses and Dissertations

Osteoarthritis (OA) is the leading cause of disability among the aging population in the United States and is frequently treated by replacing deteriorated joints with metal and plastic components. Developing better quantitative measures of movement quality to track patients longitudinally in their own homes would enable personalized treatment plans and hasten the advancement of promising new interventions. Wearable sensors and machine learning used to quantify patient movement could revolutionize the diagnosis and treatment of movement disorders. The purpose of this dissertation was to overcome technical challenges associated with the use of wearable sensors, specifically Inertial Measurement Units (IMUs), as a …

How Object Segmentation And Perceptual Grouping Emerge In Noisy Variational Autoencoders, Ben Lonnqvist, Zhengqing Wu, Michael H. Herzog

MODVIS Workshop

Many animals and humans can recognize and segment objects from their backgrounds. Whether object segmentation is necessary for object recognition has long been a topic of debate. Deep neural networks (DNNs) excel at object recognition, but not at segmentation tasks - this has led to the belief that object recognition and segmentation are separate mechanisms in visual processing. Here, however, we show evidence that in variational autoencoders (VAEs), segmentation and faithful representation of data can be interlinked. VAEs are encoder-decoder models that learn to represent independent generative factors of the data as a distribution in a very small bottleneck layer; …