Artificial Intelligence and Robotics Commons^™

Flood: A Flexible Invariant Learning Framework For Out-Of-Distribution Generalization On Graphs, Yang Liu, Xiang Ao, Fuli Feng, Yunshan Ma, Kuan Li, Tat‑Seng Chua, Qing He

Research Collection School Of Computing and Information Systems

Graph Neural Networks (GNNs) have achieved remarkable success in various domains but most of them are developed under the in-distribution assumption. Under out-of-distribution (OOD) settings, they suffer from the distribution shift between the training set and the test set and may not generalize well to the test distribution. Several methods have tried the invariance principle to improve the generalization of GNNs in OOD settings. However, in previous solutions, the graph encoder is immutable after the invariant learning and cannot be adapted to the target distribution flexibly. Confronting the distribution shift, a flexible encoder with refinement to the target distribution can …

Artificial Intelligence Frameworks To Detect And Investigate The Pathophysiology Of Spaceflight Associated Neuro-Ocular Syndrome (Sans), Joshua Ong, Ethan Waisberg, Mouayad Masalkhi, Sharif Amit Kamran, Kemper Lowry, Prithul Sarker, Nasif Zaman, Phani Paladugu, Alireza Tavakkoli, Andrew G Lee

Student Papers, Posters & Projects

Spaceflight associated neuro-ocular syndrome (SANS) is a unique phenomenon that has been observed in astronauts who have undergone long-duration spaceflight (LDSF). The syndrome is characterized by distinct imaging and clinical findings including optic disc edema, hyperopic refractive shift, posterior globe flattening, and choroidal folds. SANS serves a large barrier to planetary spaceflight such as a mission to Mars and has been noted by the National Aeronautics and Space Administration (NASA) as a high risk based on its likelihood to occur and its severity to human health and mission performance. While it is a large barrier to future spaceflight, the underlying …

Prompt-Based Tuning Of Transformer Models For Multi-Center Medical Image Segmentation Of Head And Neck Cancer, Numan Saeed, Muhammad Ridzuan, Roba Al Majzoub, Mohammad Yaqub

Computer Vision Faculty Publications

Medical image segmentation is a vital healthcare endeavor requiring precise and efficient models for appropriate diagnosis and treatment. Vision transformer (ViT)-based segmentation models have shown great performance in accomplishing this task. However, to build a powerful backbone, the self-attention block of ViT requires large-scale pre-training data. The present method of modifying pre-trained models entails updating all or some of the backbone parameters. This paper proposes a novel fine-tuning strategy for adapting a pretrained transformer-based segmentation model on data from a new medical center. This method introduces a small number of learnable parameters, termed prompts, into the input space (less than …

Understanding Political Polarization Using Language Models: A Dataset And Method, Samiran Gode, Supreeth Bare, Bhiksha Raj, Hyungon Yoo

Natural Language Processing Faculty Publications

Our paper aims to analyze political polarization in US political system using language models, and thereby help candidates make an informed decision. The availability of this information will help voters understand their candidates' views on the economy, healthcare, education, and other social issues. Our main contributions are a dataset extracted from Wikipedia that spans the past 120 years and a language model-based method that helps analyze how polarized a candidate is. Our data are divided into two parts, background information and political information about a candidate, since our hypothesis is that the political views of a candidate should be based …

On Training Neurons With Bounded Compilations, Lance Kennedy

Master of Science in Computer Science Theses

Knowledge compilation offers a formal approach to explaining and verifying the behavior of machine learning systems, such as neural networks. Unfortunately, compiling even an individual neuron into a tractable representation such as an Ordered Binary Decision Diagram (OBDD), is an NP-hard problem. In this thesis, we consider the problem of training a neuron from data, subject to the constraint that it has a compact representation as an OBDD. Our approach is based on the observation that a neuron can be compiled into an OBDD in polytime if (1) the neuron has integer weights, and (2) its aggregate weight is bounded. …

International Soft Law Governance Of Artificial Intelligence Ethics: Current Situation, Challenges And Countermeasures, Mingting Zhu, Chongli Xu

Bulletin of Chinese Academy of Sciences (Chinese Version)

Artificial intelligence (AI) technology not only rapidly empowers economic and social development, but may also trigger many ethical issues highly related to the characteristics and development of AI technology itself. The rise of international soft law in the field of AI ethical governance is almost inevitable due to its flexibility, efficiency, low application cost, ability to fill the gap in hard law, and convenience in distinguishing governance and layered response to ethical issues. Under the current situation of developed international soft law and outdated hard law in this field, faced with the governance challenge of unstable cooperation among subjects of …

How Technology May Be Used For Future Disease Predictions, Rich P. Manprisio

Journal of Applied Disciplines

Exasperated by the ongoing global pandemic, the healthcare system is grappling with the formidable challenges posed by proper and effective disease treatments. Nevertheless, amidst these growing difficulties, the healthcare field has witnessed significant technological advancements, offering promising avenues for disease prediction. Notably, a positive correlation exists between the utilization of technologies and their potential to serve as valuable tools for disease prediction. As our reliance on technological sophistication continues progressing, current research highlights numerous viable options to augment the healthcare sector. This review explores the current state of utilizing technologies and their potential to enhance healthcare, shedding light on their …

Using Machine Learning Techniques To Model Encoder/Decoder Pair For Non-Invasive Electroencephalographic Wireless Signal Transmission, Ernst Fanfan

Master of Science in Computer Science Theses

This study investigated the application and enhancement of Non-Invasive Brain-Computer Interfaces (NI-BCIs), focused on enhancing the efficiency and effectiveness of this technology for individuals with severe physical limitations. The core research goal was to improve current limitations associated with wires, noise, and invasive procedures often associated with BCI technology. The key discussed solution involves developing an optimized Encoder/Decoder (E/D) pair using machine learning techniques, particularly those borrowed from Generative Adversarial Networks (GAN) and other Deep Neural Networks, to minimize data transmission and ensure robustness against data degradation. The study highlighted the crucial role of machine learning in self-adjusting and isolating …

Enhancing Video-Based Learning Using Knowledge Tracing: Personalizing Students’ Learning Experience With Orbits, Shady Shehata, David Santandreu, Philip Purnell, Mark Thompson

Natural Language Processing Faculty Publications

As the world regains its footing following the COVID-19 pandemic, academia is striving to consolidate the gains made in students’ education experience. New technologies such as video-based learning have shown some early improvement in student learning and engagement. In this paper, we present ORBITS predictive engine at YOURIKA company, a video-based student support platform powered by knowledge tracing. In an exploratory case study of one master’s level Speech Processing course at the Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI) in Abu Dhabi, half the students used the system while the other half did not. Student qualitative feedback was universally …

Ways To Participate In Ongoing Regulation Around Artificial Intelligence Ethics In The United States, Wilhelmina Randtke

University Libraries: Faculty Presentations

In January 2021, the US passed the National Artificial Intelligence Initiative Act of 2020. The goal is a cohesive federal AI initiative, and part of that is safety, ethics, and transparency. The act includes funding appropriations for 2021-2025, and roll out takes place over that time. In implementing this law, there is recent and ongoing activity to regulate AI in the US. Regular calls for public participation go out to the public on www.federalregister.gov in the form of open ended questions on which input is requested, and feedback on reports or action plans.

The linked data community is uniquely positioned …

On Teaching Multi-Criteria Decision Making With A Robot Assistant, Chen Zhang, Hakan Saraoglu, David A. Louton

Information Systems and Analytics Department Faculty Conference Proceedings

We propose a system and method for a robot assistant for teaching multi-attribute decision making (MCDM). Through questions and answers in natural language, the robot assistant learns the user’s preferences on multiple criteria involving a selection decision and makes recommendations using data on each criterion and the learned user preferences. It will include a use-case demonstration where NAO the robot will assist a human in forming a simple portfolio of mutual funds. Presenters will illustrate the architecture of the robot assisted MCDM and describe a method that is extensively used to structure complex decision problems and has been applied to …

Library Copyright Alliance Principles For Copyright And Artificial Intelligence, Library Copyright Alliance, American Library Association, Association Of Research Libraries

Copyright, Fair Use, Scholarly Communication, etc.

Library Copyright Alliance principles for copyright and artificial intelligence, July 10, 2023.

Towards Enabling Haptic Communications Over 6g: Issues And Challenges, Muhammad Awais, Fasih Ullah Khan, Muhammad Zafar, Muhammad Mudassar, Muhammad Zaigham Zaheer, Khalid Mehmood Cheema, Muhammad Kamran, Woo Sung Jung

Computer Vision Faculty Publications

This research paper provides a comprehensive overview of the challenges and potential solutions related to enabling haptic communication over the Tactile Internet in the context of 6G networks. The increasing demand for multimedia services and device proliferation has resulted in limited radio resources, posing challenges in their efficient allocation for Device-to-Device (D2D)-assisted haptic communications. Achieving ultra-low latency, security, and energy efficiency are crucial requirements for enabling haptic communication over TI. The paper explores various methodologies, technologies, and frameworks that can facilitate haptic communication, including backscatter communications (BsC), non-orthogonal multiple access (NOMA), and software-defined networks. Additionally, it discusses the potential of …

Face Readers: The Frontier Of Computer Vision And Math Learning, Beverly Woolf, Margrit Betke, Hao Yu, Sarah Adel Bargal, Ivan Arroyo, John J. Magee Iv, Danielle Allessio, William Rebelsky

Computer Science

The future of AI-assisted individualized learning includes computer vision to inform intelligent tutors and teachers about student affect, motivation and performance. Facial expression recognition is essential in recognizing subtle differences when students ask for hints or fail to solve problems. Facial features and classification labels enable intelligent tutors to predict students’ performance and recommend activities. Videos can capture students’ faces and model their effort and progress; machine learning classifiers can support intelligent tutors to provide interventions. One goal of this research is to support deep dives by teachers to identify students’ individual needs through facial expression and to provide immediate …

Case Study: The Impact Of Emerging Technologies On Cybersecurity Education And Workforces, Austin Cusak

Journal of Cybersecurity Education, Research and Practice

A qualitative case study focused on understanding what steps are needed to prepare the cybersecurity workforces of 2026-2028 to work with and against emerging technologies such as Artificial Intelligence and Machine Learning. Conducted through a workshop held in two parts at a cybersecurity education conference, findings came both from a semi-structured interview with a panel of experts as well as small workgroups of professionals answering seven scenario-based questions. Data was thematically analyzed, with major findings emerging about the need to refocus cybersecurity STEM at the middle school level with problem-based learning, the disconnects between workforce operations and cybersecurity operators, the …

Target-Based Offensive Language Identification, Marcos Zampieri, Skye Morgan, Kai North, Tharindu Ranasinghe, Austin Simmons, Paridhi Khandelwal, Sara Rosenthal, Preslav Nakov

Natural Language Processing Faculty Publications

We present TBO, a new dataset for Target-based Offensive language identification. TBO contains post-level annotations regarding the harmfulness of an offensive post and token-level annotations comprising of the target and the offensive argument expression. Popular offensive language identification datasets for social media focus on annotation taxonomies only at the post level and more recently, some datasets have been released that feature only token-level annotations. TBO is an important resource that bridges the gap between post-level and token-level annotation datasets by introducing a single comprehensive unified annotation taxonomy. We use the TBO taxonomy to annotate post-level and token-level offensive language on …

Analysis Of Predictive Performance And Reliability Of Classifiers For Quality Assessment Of Medical Evidence Revealed Important Variation By Medical Area, Simon Šuster, Timothy Baldwin, Karin Verspoor

Natural Language Processing Faculty Publications

Objectives: A major obstacle in deployment of models for automated quality assessment is their reliability. To analyze their calibration and selective classification performance. Study Design and Setting: We examine two systems for assessing the quality of medical evidence, EvidenceGRADEr and RobotReviewer, both developed from Cochrane Database of Systematic Reviews (CDSR) to measure strength of bodies of evidence and risk of bias (RoB) of individual studies, respectively. We report their calibration error and Brier scores, present their reliability diagrams, and analyze the risk–coverage trade-off in selective classification. Results: The models are reasonably well calibrated on most quality criteria (expected calibration error …

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 …

Linear Classifier: An Often-Forgotten Baseline For Text Classification, Yu Chen Lin, Si An Chen, Jie Jyun Liu, Chih Jen Lin

Machine Learning Faculty Publications

Large-scale pre-trained language models such as BERT are popular solutions for text classification. Due to the superior performance of these advanced methods, nowadays, people often directly train them for a few epochs and deploy the obtained model. In this opinion paper, we point out that this way may only sometimes get satisfactory results. We argue the importance of running a simple baseline like linear classifiers on bag-of-words features along with advanced methods. First, for many text data, linear methods show competitive performance, high efficiency, and robustness. Second, advanced models such as BERT may only achieve the best results if properly …

Team Thesyllogist At Semeval-2023 Task 3: Language-Agnostic Framing Detection In Multi-Lingual Online News: A Zero-Shot Transfer Approach, Osama Mohammed Afzal, Preslav Nakov

Natural Language Processing Faculty Publications

We describe our system for SemEval-2022 Task 3 subtask 2 which on detecting the frames used in a news article in a multi-lingual setup. We propose a multi-lingual approach based on machine translation of the input, followed by an English prediction model. Our system demonstrated good zero-shot transfer capability, achieving micro-F1 scores of 53% for Greek (4th on the leaderboard) and 56.1% for Georgian (3rd on the leaderboard), without any prior training on translated data for these languages. Moreover, our system achieved comparable performance on seven other languages, including German, English, French, Russian, Italian, Polish, and Spanish. Our results demonstrate …