Digital Commons Network^™

Stereotypes And Language Models: Understanding How Language Models Encode Stereotypes, Debiasing Language Models, And Examining How Stereotypes Affect Conversations, Brian C. Wang

Computer Science Senior Theses

This thesis describes a variety of approaches in examining how language models encode stereotypes (understanding stereotypes from a model point-of-view), debiasing language models, and using language models to understand how stereotypes affect conversations (understanding stereotypes from a conversational point-of-view). We present a novel approach for textual clues analysis that makes language models more interpretable, combining the understanding of what stereotypes the internal structures of language models have encoded during their initial training (via attention-based analysis) and understanding what textual clues are most relevant to identifying stereotypes for models trained to detect stereotypes (via SHAP-based analysis). We find that different pre-trained …

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, …

Counterfactual Replacement Analysis For Interpretation Of Blackbox Sexism Classification Models, Anders Knospe

Computer Science Senior Theses

This paper describes the AKD team’s system designed for SemEval-2023 Task 10: Explainable Detection of Online Sexism (EDOS). We implement a simple fine-tuned GPT-3 model, ranking 26 on the leaderboard for task A. We also discuss different approaches to interpretability in the context of critiquing the EDOS task’s sub-category oriented approach. Finally, we propose counterfactual replacement analysis, a novel prototype technique for approaching explainability.

Dialogs Re-Enacted Across Languages, Version 2, Nigel G. Ward, Jonathan E. Avila, Emilia Rivas, Divette Marco

Departmental Technical Reports (CS)

To support machine learning of cross-language prosodic mappings and other ways to improve speech-to-speech translation, we present a protocol for collecting closely matched pairs of utterances across languages, a description of the resulting data collection and its public release, and some observations and musings. This report is intended for:

• people using this corpus
• people extending this corpus
• people designing similar collections of bilingual dialog data.

Change Notes. This version supersedes UTEP-CS-22-108. There is some new information and numerous clarifications, mostly arising from our experiences diversifying our corpus and helping a vendor to use this protocol.

Automatic Scoring Of Speeded Interpersonal Assessment Center Exercises Via Machine Learning: Initial Psychometric Evidence And Practical Guidelines, Louis Hickman, Christoph N. Herde, Filip Lievens, Louis Tay

Research Collection Lee Kong Chian School Of Business

Assessment center (AC) exercises such as role-plays have established themselves as valuable approaches for obtaining insights into interpersonal behavior, but they are often considered the “Rolls Royce” of personnel assessment due to their high costs. The observation and rating process comprises a substantial part of these costs. In an exploratory case study, we capitalize on recent advances in natural language processing (NLP) by developing NLP-based machine learning (ML) models to investigate the possibility of automatically scoring AC exercises. First, we compared the convergent-related validity and contamination with word count of ML scores based on models that used different NLP methods …

Exploiting Association Rules Mining To Inform The Use Of Non-Manual Features In Sign Language Processing, Robert G. Smith

Other Resources

In recent years, the use of virtual assistants and voice user interfaces has become a latent part of modern living. Unseen to the user are the various artificial intelligence and natural language processing technologies, the vast datasets, and the linguistic insights that underpin such tools. The technologies supporting them have chiefly targeted widely used spoken languages, leaving sign language users at a disadvantage. One important reason why sign languages are unsupported by such tools is a requirement of the underpinning technologies for a comprehensive description of the language. Sign language processing technologies endeavour to bridge this technology inequality.

Recent approaches …

Neural Tabula Rasa: Foundations For Realistic Memories And Learning, Patrick R. Perrine

Master's Theses

Understanding how neural systems perform memorization and inductive learning tasks are of key interest in the field of computational neuroscience. Similarly, inductive learning tasks are the focus within the field of machine learning, which has seen rapid growth and innovation utilizing feedforward neural networks. However, there have also been concerns regarding the precipitous nature of such efforts, specifically in the area of deep learning. As a result, we revisit the foundation of the artificial neural network to better incorporate current knowledge of the brain from computational neuroscience. More specifically, a random graph was chosen to model a neural system. This …

Stream-Evolving Bot Detection Framework Using Graph-Based And Feature-Based Approaches For Identifying Social Bots On Twitter, Eiman Alothali

Dissertations

This dissertation focuses on the problem of evolving social bots in online social networks, particularly Twitter. Such accounts spread misinformation and inflate social network content to mislead the masses. The main objective of this dissertation is to propose a stream-based evolving bot detection framework (SEBD), which was constructed using both graph- and feature-based models. It was built using Python, a real-time streaming engine (Apache Kafka version 3.2), and our pretrained model (bot multi-view graph attention network (Bot-MGAT)). The feature-based model was used to identify predictive features for bot detection and evaluate the SEBD predictions. The graph-based model was used to …

Position-Guided Text Prompt For Vision-Language Pre-Training, Alex Jinpeng Wang, Pan Zhou, Mike Zheng Shou, Yan Shuicheng

Research Collection School Of Computing and Information Systems

Vision-Language Pre-Training (VLP) has shown promising capabilities to align image and text pairs, facilitating a broad variety of cross-modal learning tasks. However, we observe that VLP models often lack the visual grounding/localization capability which is critical for many downstream tasks such as visual reasoning. In this work, we propose a novel Position-guided Text Prompt (PTP) paradigm to enhance the visual grounding ability of cross-modal models trained with VLP. Specifically, in the VLP phase, PTP divides the image into N x N blocks, and identifies the objects in each block through the widely used object detector in VLP. It then reformulates …

Multi-Head Attention Graph Convolutional Network Model: End-To-End Entity And Relation Joint Extraction Based On Multi-Head Attention Graph Convolutional Network, Zhihua Tao, Chunping Ouyang, Yongbin Liu, Tonglee Chung, Yixin Cao

Research Collection School Of Computing and Information Systems

At present, the entity and relation joint extraction task has attracted more and more scholars' attention in the field of natural language processing (NLP). However, most of their methods rely on NLP tools to construct dependency trees to obtain sentence structure information. The adjacency matrix constructed by the dependency tree can convey syntactic information. Dependency trees obtained through NLP tools are too dependent on the tools and may not be very accurate in contextual semantic description. At the same time, a large amount of irrelevant information will cause redundancy. This paper presents a novel end-to-end entity and relation joint extraction …

Evaluating Neural Networks As Cognitive Models For Learning Quasi-Regularities In Language, Xiaomeng Ma

Dissertations, Theses, and Capstone Projects

Many aspects of language can be categorized as quasi-regular: the relationship between the inputs and outputs is systematic but allows many exceptions. Common domains that contain quasi-regularity include morphological inflection and grapheme-phoneme mapping. How humans process quasi-regularity has been debated for decades. This thesis implemented modern neural network models, transformer models, on two tasks: English past tense inflection and Chinese character naming, to investigate how transformer models perform quasi-regularity tasks. This thesis focuses on investigating to what extent the models' performances can represent human behavior. The results show that the transformers' performance is very similar to human behavior in many …

Mapping Programs To Equations, Hessamaldin Mohammadi

Dissertations

Extracting the function of a program from a static analysis of its source code is a valuable capability in software engineering; at a time when there is increasing talk of using AI (Artificial Intelligence) to generate software from natural language specifications, it becomes increasingly important to determine the exact function of software as written, to figure out what AI has understood the natural language specification to mean. For all its criticality, the ability to derive the domain-to-range function of a program has proved to be an elusive goal, due primarily to the difficulty of deriving the function of iterative statements. …

Deep Hybrid Modeling Of Neuronal Dynamics Using Generative Adversarial Networks, Soheil Saghafi

Dissertations

Mechanistic modeling and machine learning methods are powerful techniques for approximating biological systems and making accurate predictions from data. However, when used in isolation these approaches suffer from distinct shortcomings: model and parameter uncertainty limit mechanistic modeling, whereas machine learning methods disregard the underlying biophysical mechanisms. This dissertation constructs Deep Hybrid Models that address these shortcomings by combining deep learning with mechanistic modeling. In particular, this dissertation uses Generative Adversarial Networks (GANs) to provide an inverse mapping of data to mechanistic models and identifies the distributions of mechanistic model parameters coherent to the data.

Chapter 1 provides background information on …

Connecting Linguistic Expressions And Pain Relief Through Transformer Model Construction And Analysis, Sarah M. Chacko

Computer Science Senior Theses

Chronic pain is a widespread problem that significantly impacts quality of life. Overprescription and abuse of pain medication continues to be a major public health issue and can further burden patients due to a fragmented health care system. Previous research has suggested a possible psychological basis to pain and the potential for safer, non-pharmacological alternatives for pain relief. This project leverages language models to study chronic pain development and relief through psychological treatments, which will be assessed through responses to post-treatment interviews. A transformer-based natural language processing model is employed to identify connections between language expressions and pain on a …

Investigating English-Language Dialect-Adjusted Models, Samiha Datta

Computer Science Senior Theses

This thesis describes several approaches to better understand how large language models interpret different dialects of the English language. Our goal is to consider multiple contexts of textual data and to analyze how English-language dialects are realized in them, as well as how a variety of machine learning techniques handle these differences. We focus on two genres of text data: news and social media. In the news context, we establish a dataset covering news articles from five countries and four US states and consider language modeling analysis, topic and sentiment distributions, and manual analysis before performing nine experiments and evaluating …

Neutrosophic Mcdm Methodology For Risk Assessment Of Autonomous Underwater Vehicles, Shimaa S. Mohamed, Ahmed Abdel-Monem, Alshaimaa A. Tantawy

Neutrosophic Systems with Applications

Due to its usefulness in several industries and the military, researchers have concentrated on developing autonomous underwater vehicles (AUVs). However, AUV navigation continues to be a difficult challenge to solve owing to the variety of underwater settings. The usage of AUVs, or autonomous underwater vehicles, is not without dangers like malfunction, ecological risks, loss of communications, cybersecurity risks, collisions, and others. There are many criteria to assess these risks technical, operational, economic, and regulatory. So, the methods of multi-criteria decision-making (MCDM) is used to deal with these various criteria. The analytical hierarchy process (AHP) method is an MCDM methodology, that …

A Machine Learning Model To Predict The Asd Traits And Challenging Behaviors In Children, Biswaranjan Senapati

Theses and Dissertations

A neurological disorder, along with several behavioral issues, may be to blame for a child's subpar performance in the academic journey (such as anxiety, depression, learning disorders, and irritability). These symptoms can be used to diagnose children with ASD, and supervised machine learning models can help differentiate between ASD traits and other behavioral traits in children, so that a correct diagnosis can be made. Autism usually shows up in a child within the first two years of life, but it can happen at any time. This research aims to create a model for predicting ASD traits and challenging behaviors in …

Progress In A New Visualization Strategy For Ml Models, Alex Wissing, Brian Ricks, Robin Gandhi, Yonas Kassa, Akshay Kale

Information Systems and Quantitative Analysis Faculty Proceedings & Presentations

No abstract provided.

Handling Realistic Label Noise In Bert Text Classification, Maha Tufail Agro, Hanan Al Darmaki

Natural Language Processing Faculty Publications

Label noise refers to errors in training labels caused by cheap data annotation methods, such as web scraping or crowd-sourcing, which can be detrimental to the performance of supervised classifiers. Several methods have been proposed to counteract the effect of random label noise in supervised classification, and some studies have shown that BERT is already robust against high rates of randomly injected label noise. However, real label noise is not random; rather, it is often correlated with input features or other annotator-specific factors. In this paper, we evaluate BERT in the presence of two types of realistic label noise: feature-dependent …

Detecting Ai Generated Text Using Neural Networks, Jesus Guerrero

Masters Theses (Archived)

For humans, distinguishing machine generated text from human written text is men- tally taxing and slow. NLP models have been created to do this more effectively and faster. But, what if some adversarial changes have been added to the machine generated text? This thesis discusses this issue and text detectors in general.
/="/">
/="/">The primary goal of this thesis is to describe the current state of text detectors in research and to discuss a key adversarial issue in modern NLP transformers. To describe the current state of text detectors a Systematic Literature Review was done on 50 relevant papers …

Ptmtorrent: A Dataset For Mining Open-Source Pre-Trained Model Packages, Wenxin Jiang, Nicholas Synovic, Purvish Jajal, Taylor R. Schorlemmer, Arav Tewari, Bhavesh Pareek, George K. Thiruvathukal, James C. Davis

Computer Science: Faculty Publications and Other Works

Due to the cost of developing and training deep learning models from scratch, machine learning engineers have begun to reuse pre-trained models (PTMs) and fine-tune them for downstream tasks. PTM registries known as “model hubs” support engineers in distributing and reusing deep learning models. PTM packages include pre-trained weights, documentation, model architectures, datasets, and metadata. Mining the information in PTM packages will enable the discovery of engineering phenomena and tools to support software engineers. However, accessing this information is difficult — there are many PTM registries, and both the registries and the individual packages may have rate limiting for accessing …

Generative And Pseudo-Relevant Feedback For Sparse, Dense And Learned Sparse Retrieval, Iain Mackie, Shubham Chatterjee, Jeffrey Dalton

Computer Science Faculty Research & Creative Works

Pseudo-relevance feedback (PRF) is a classical approach to address lexical mismatch by enriching the query using first-pass retrieval. Moreover, recent work on generative-relevance feedback (GRF) shows that query expansion models using text generated from large language models can improve sparse retrieval without depending on first-pass retrieval effectiveness. This work extends GRF to dense and learned sparse retrieval paradigms with experiments over six standard document ranking benchmarks. We find that GRF improves over comparable PRF techniques by around 10% on both precision and recall-oriented measures. Nonetheless, query analysis shows that GRF and PRF have contrasting benefits, with GRF providing external context …

Du Undergraduate Showcase: Research, Scholarship, And Creative Works, Caitlyn Aldersea, Justin Bravo, Sam Allen, Anna Block, Connor Block, Emma Buechler, Maria De Los Angeles Bustillos, Arianna Carlson, William Christensen, Olivia Kachulis, Noah Craver, Kate Dillon, Muskan Fatima, Angel Fernandes, Emma Finch, Colleen Cassidy, Amy Fishman, Andrea Francis, Stacia Fritz, Simran Gill, Emma Gries, Rylie Hansen, Shannon Powers, Jacqueline Martinez, Zachary Harker, Ashley Hasty, Mykaela Tanino-Springsteen, Kathleen Hopps, Adelaide Kerenick, Colin Kleckner, Ci Koehring, Elijah Kruger, Braden Krumholz, Maddie Leake, Lyneé Alves, Seraphina Loukas, Yatzari Lozano Vazquez, Haley Maki, Emily Martinez, Sierra Mckinney, Audrey Mitchell, Kipling Newman, Audrey Ng, Megan Lucyshyn, Andrew Nguyen, Stevie Ostman, Casandra Pearson, Alexandra Penney, Julia Gielczynski, Tyler Ball, Anna Rini, Christina Rorres, Simon Ruland, Helayna Schafer, Emma Sellers, Sarah Schuller, Claire Shaver, Kevin Summers, Isabella Shaw, Madison Sinar, Claudia Pena, Apshara Siwakoti, Carter Sorensen, Madi Sousa, Anna Sparling, Alexandra Revier, Brandon Thierry, Dylan Tyree, Maggie Williams, Lauren Wols

DU Undergraduate Research Journal Archive

DU Undergraduate Showcase: Research, Scholarship, and Creative Works

Topic Modeling And Future Prediction Of Aid Data In Development Studies Using Lda And Bert, Uttamasha Anjally Oyshi

Theses and Dissertations

This thesis presents a study on topic modeling and future prediction of aid data in development studies using LDA and BERT. The goal of this study is to explore the aid data from four sectors: Government and civil society, Government and civil society general, Conflict prevention and resolution, peace and security, and Women, and to identify the latent topics that are present in the data. The LDA and BERT algorithms were used for the topic modeling, and coherence scores were computed to evaluate the quality of the models. The results of the study show that the LDA and BERT models …

Fair Enough: Standardizing Evaluation And Model Selection For Fairness Research In Nlp, Xudong Han, Timothy Baldwin, Trevor Cohn

Natural Language Processing Faculty Publications

Modern NLP systems exhibit a range of biases, which a growing literature on model debiasing attempts to correct. However, current progress is hampered by a plurality of definitions of bias, means of quantification, and oftentimes vague relation between debiasing algorithms and theoretical measures of bias. This paper seeks to clarify the current situation and plot a course for meaningful progress in fair learning, with two key contributions: (1) making clear inter-relations among the current gamut of methods, and their relation to fairness theory; and (2) addressing the practical problem of model selection, which involves a trade-off between fairness and accuracy …

Music Mentor, Jacob Webb

Honors College Theses

Extra-curricular learning is on the rise, and many are interested in expanding their current knowledge by utilizing the recent increase in educational technology. While many forms of educational technology exist, there are few interactive and engaging platforms that teach music theory. Apps such as Perfect Ear and MyMusicTheory are great for becoming familiar with reading music and recognizing pitches, however, they often become dry with repetition and repeated tasks. By combining existing technologies that can complete real time conversions from raw audio to MIDI, our goal was to gather information such as harmonies, key and compatible chords from the user’s …

Enhancing Health Tweet Classification: An Evaluation Of Transformer-Based Models For Comprehensive Analysis, Foram Pankajbhai Patel

Computer Science and Engineering Theses - Archive

The task of health tweet classification entails identifying whether a given tweet is health-related or not. While existing research in this area has made significant progress in classifying tweets into specific sub-domains of health, such as mental health, COVID-19, or specific diseases, there is a need for a more comprehensive approach that considers a broader range of health-related topics. This thesis addresses this need by proposing a diverse and comprehensive dataset that includes various existing health-related datasets, data collected through a keyword-based approach, and manually annotated data. However, the use of health-related keywords in a figurative or non-health context poses …

Toward Digital Phenotyping: Human Activity Representation For Embodied Cognition Assessment, Mohammad Zakizadehghariehali

Computer Science and Engineering Dissertations - Archive

Cognition is the mental process of acquiring knowledge and understanding through thought, experience and senses. Based on Embodied Cognition theory, physical activities are an important manifestation of cognitive functions. As a result, they can be employed to both assess and train cognitive skills. In order to assess various cognitive measures, the ATEC system has been proposed. It consists of physical exercises with different variations and difficulty levels, designed to provide assessment of executive and motor functions. This thesis focuses on obtaining human activity representation from recorded videos of ATEC tasks in order to automatically assess embodied cognition performance. Representation learning …

Fake News Detection Using Narrative Content And Discourse, Hongmin Kim

Boise State University Theses and Dissertations

With the growth of modern technology, we are living in a world where anyone can share news with the tap of a finger. The simplified process of news sharing has brought an inundation of information on the Internet, along with a vast amount of fake news. Researchers have been working to understand the characteristics of fake news, in order to accurately identify them through automated text analysis.

In this thesis, we propose the Narrative Content and Narrative Discourse features brought from the ideas by van Laer et al., Berger et al., and Aleti et al. We performed various experiments including …

Predicting High-Cap Tech Stock Polarity: A Combined Approach Using Support Vector Machines And Bidirectional Encoders From Transformers, Ian L. Grisham

Electronic Theses and Dissertations

The abundance, accessibility, and scale of data have engendered an era where machine learning can quickly and accurately solve complex problems, identify complicated patterns, and uncover intricate trends. One research area where many have applied these techniques is the stock market. Yet, financial domains are influenced by many factors and are notoriously difficult to predict due to their volatile and multivariate behavior. However, the literature indicates that public sentiment data may exhibit significant predictive qualities and improve a model’s ability to predict intricate trends. In this study, momentum SVM classification accuracy was compared between datasets that did and did not …