Open Access. Powered by Scholars. Published by Universities.®

Digital Commons Network™

Open Access. Powered by Scholars. Published by Universities.®

2024

Discipline
Institution
Keyword
Publication
Publication Type
File Type

Articles 24931 - 24960 of 27097

Full-Text Articles in Entire DC Network

Improving The Robustness Of Neural Networks To Adversarial Patch Attacks Using Masking And Attribution Analysis, Atandra Mahalder Jan 2024

Improving The Robustness Of Neural Networks To Adversarial Patch Attacks Using Masking And Attribution Analysis, Atandra Mahalder

Honors Undergraduate Theses

Computer vision algorithms, including image classifiers and object detectors, play a pivotal role in various cyber-physical systems, spanning from facial recognition to self-driving vehicles and security surveillance. However, the emergence of real-world adversarial patches, which can be as simple as stickers, poses a significant threat to the reliability of AI models utilized within these systems. To address this challenge, several defense mechanisms such as PatchGuard, Minority Report, and (De)Randomized Smoothing have been proposed to enhance the resilience of AI models against such attacks. In this thesis, we introduce a novel framework that integrates masking with attribution analysis to robustify AI …

Go to article

Estimating Modeling Parameters For Covid-19 Spread On Campus, Aviel S. Crigger Jan 2024

Estimating Modeling Parameters For Covid-19 Spread On Campus, Aviel S. Crigger

Honors Undergraduate Theses

Understanding the true burden of community transmission of communicable diseases like COVID-19 is crucial for effective public health response. Clinical cases, while important, only represent a fraction of the actual disease prevalence within a population. In this thesis, we investigate methods to estimate parameters that link clinical cases to the true disease prevalence using a modified compartmental model known as SICR (Susceptible, Infected, Cases, Recovered). We employ Bayesian inference and ensemble Markov chain Monte Carlo (MCMC) simulations to analyze clinical case data provided by the University of Central Florida Health Center from 2020 to 2022. Our goal is to estimate …

Go to article

Tracking The Reaction Networks Of Criegee Intermediates Formed In The Ozone-Assisted Oxidation Reactions Of C5 Acyclic And Endocyclic Alkenes, Arden M. Floyd Jan 2024

Tracking The Reaction Networks Of Criegee Intermediates Formed In The Ozone-Assisted Oxidation Reactions Of C5 Acyclic And Endocyclic Alkenes, Arden M. Floyd

Honors Undergraduate Theses

Uni- and bi-molecular reactions involving Criegee intermediates (CIs), characteristic to alkene-ozone reactions, play an important role in synthetic organic, atmospheric, and combustion chemistry. However, difficulty in detecting key intermediates has prevented a thorough understanding of the details of the ozonolysis mechanism. To gain new insights into how alkene structure affects the CI network, the ozone assisted oxidation reactions of trans-2-pentene and cyclopentene are investigated between 320 K and 760 K in an atmospheric pressure jet stirred reactor (JSR). Molecular-beam mass spectrometry in conjunction with single photon tunable synchrotron VUV radiation is used to identify elusive intermediates by means of …

Go to article

The Effects Of Nanobubbles For Gene Delivery For Osteoporosis Treatment, Abel Córdova Flores Jan 2024

The Effects Of Nanobubbles For Gene Delivery For Osteoporosis Treatment, Abel Córdova Flores

Honors Undergraduate Theses

In recent years, the field of nanomedicine has provided promising alternatives to traditional ways of gene and drug delivery, such as nanobubbles instead of viral vectors. Manufacturing these nanomaterials is cost-effective and tends to be safer than previous forms of treatment for clinical use. In the realm of osteoporosis, a disease based around a higher rate of bone resorption to bone formation, leads to weak and brittle bones, leaving sufferers at risk for fractures, which are sudden and unexpected. These fractures are estimated to cost USD 22 Billion annually in the US alone. Reports show a significant effect on the …

Go to article

Shades Of Justice: Exploring Colorism In The Hispanic Community And Its Legal Battle For Equity, Christel A. Infante Jan 2024

Shades Of Justice: Exploring Colorism In The Hispanic Community And Its Legal Battle For Equity, Christel A. Infante

Honors Undergraduate Theses

This thesis focuses on the racial disparity within the Hispanic and Latinx communities as injustices exist within the community and the workplace. Racial disparities in the United States have been a persistent and deeply rooted issue that has plagued the nation for centuries. Despite significant progress in civil rights and anti-discrimination legislation, disparities in areas such as education, employment, and criminal justice persist. Understanding the factors contributing to these disparities is essential for addressing systemic inequalities and fostering a more just society. The analysis of this thesis primarily focuses on the cases and ramifications of Hispanic persons within the workplace, …

Go to article

Music Recommendation Using Exemplars And Contrastive Learning, Tina Tran Jan 2024

Music Recommendation Using Exemplars And Contrastive Learning, Tina Tran

Honors Undergraduate Theses

The popularity of AI audio applications is growing, it is used in chatbots, automated voice translation, virtual assistants, and text-to-speech translation. Audio classification is crucial in today’s world with a growing need to sort and classify millions of existing audio data with increasing amounts of new data uploaded over time. In the area of classification lies the difficult and lucrative problem of music recommendation. Research in music recommendation has trended over time towards collaborative-based approaches utilizing large amounts of user data. These approaches tend to deal with the cold-start problem of insufficient data and are costly to train. We look …

Go to article

Multicolor Bipartite Ramsey Number Of Double Stars, Gregory M. Decamillis Jan 2024

Multicolor Bipartite Ramsey Number Of Double Stars, Gregory M. Decamillis

Honors Undergraduate Theses

The core idea of Ramsey theory is that complete disorder is impossible. Given a large structure, no matter how complex it is, we can always find a smaller substructure that has some sort of order. For positive integers $n, m$, the double star $S(n,m)$ is the graph consisting of the disjoint union of two stars $K_{1,n}$ and $K_{1,m}$ together with an edge joining their centers. The $k$-color bipartite Ramsey number of $ S(n,m)$, denoted by $r_{bip}(S(n,m);k)$, is the smallest integer $N$ such that, in any $k$-coloring of the edges of the complete bipartite graph $K_{N,N}$, there is a monochromatic copy …

Go to article

L3dml: Facilitating Geo-Distributed Machine Learning In Network Layer, Xindi Hou, Shuai Gao, Ningchun Liu, Fangtao Yao, Bo Lei, Hongke Zhang, Sajal K. Das Jan 2024

L3dml: Facilitating Geo-Distributed Machine Learning In Network Layer, Xindi Hou, Shuai Gao, Ningchun Liu, Fangtao Yao, Bo Lei, Hongke Zhang, Sajal K. Das

Computer Science Faculty Research & Creative Works

Geo-Distributed Machine Learning (GDML) aims to train large-scale machine learning models across geographically dispersed datacenters. However, the performance of GDML systems is constrained by the limited Wide Area Network (WAN) bandwidth and the presence of the straggler problem. Existing GDML designs often show contradictory effects in addressing these challenges, while in-network computing attempts are typically restricted to single datacenter environments rather than the more complex GDML scenarios. To overcome these limitations, this paper proposes L3DML to facilitate GDML using the P4-based Software-defined Network (SDN). Our approach incorporates three key innovations. Firstly, we introduce a novel network addressing scheme that enables …

Go to article

Message From The Phd Dissertation Showcase Chairs, Panos K. Chrysanthis, Sanjay Madria Jan 2024

Message From The Phd Dissertation Showcase Chairs, Panos K. Chrysanthis, Sanjay Madria

Computer Science Faculty Research & Creative Works

No abstract provided.

Go to article

Smart Connected Farms And Networked Farmers To Improve Crop Production, Sustainability And Profitability, Asheesh K. Singh, Behzad J. Balabaygloo, Barituka Bekee, Samuel W. Blair, Suzanne Fey, Fateme Fotouhi, Ashish Gupta, Amit Jha, Jorge C. Martinez-Palomares, Kevin Menke, Aaron Prestholt, Vishesh K. Tanwar, Xu Tao, Anusha Vangala, Matthew E. Carroll, Sajal K. Das, Guilherme Depaula Jan 2024

Smart Connected Farms And Networked Farmers To Improve Crop Production, Sustainability And Profitability, Asheesh K. Singh, Behzad J. Balabaygloo, Barituka Bekee, Samuel W. Blair, Suzanne Fey, Fateme Fotouhi, Ashish Gupta, Amit Jha, Jorge C. Martinez-Palomares, Kevin Menke, Aaron Prestholt, Vishesh K. Tanwar, Xu Tao, Anusha Vangala, Matthew E. Carroll, Sajal K. Das, Guilherme Depaula

Computer Science Faculty Research & Creative Works

To meet the grand challenges of agricultural production including climate change impacts on crop production, a tight integration of social science, technology and agriculture experts including farmers are needed. Rapid advances in information and communication technology, precision agriculture and data analytics, are creating a perfect opportunity for the creation of smart connected farms (SCFs) and networked farmers. a network and coordinated farmer network provide unique advantages to farmers to enhance farm production and profitability, while tackling adverse climate events. the aim of this article is to provide a comprehensive overview of the state of the art in SCF including the …

Go to article

Protecting Activity Sensing Data Privacy Using Hierarchical Information Dissociation, Guangjing Wang, Hanqing Guo, Yuanda Wang, Bocheng Chen, Ce Zhou, Qiben Yan Jan 2024

Protecting Activity Sensing Data Privacy Using Hierarchical Information Dissociation, Guangjing Wang, Hanqing Guo, Yuanda Wang, Bocheng Chen, Ce Zhou, Qiben Yan

Computer Science Faculty Research & Creative Works

Smartphones and wearable devices have been integrated into our daily lives, offering personalized services. However, many apps become overprivileged as their collected sensing data contains unnecessary sensitive information. For example, mobile sensing data could reveal private attributes (e.g., gender and age) and unintended sensitive features (e.g., hand gestures when entering passwords). To prevent sensitive information leakage, existing methods must obtain private labels and users need to specify privacy policies. However, they only achieve limited control over information disclosure. In this work, we present Hippo to dissociate hierarchical information including private metadata and multi-grained activity information from the sensing data. Hippo …

Go to article

Interlude: Interactions Between Labeled And Unlabeled Data To Enhance Semi-Supervised Learning, Zhe Huang, Xiaowei Yu, Dajiang Zhu, Michael C. Hughes Jan 2024

Interlude: Interactions Between Labeled And Unlabeled Data To Enhance Semi-Supervised Learning, Zhe Huang, Xiaowei Yu, Dajiang Zhu, Michael C. Hughes

Computer Science Faculty Research & Creative Works

Semi-supervised learning (SSL) seeks to enhance task performance by training on both labeled and unlabeled data. Mainstream SSL image classification methods mostly optimize a loss that additively combines a supervised classification objective with a regularization term derived solely from unlabeled data. This formulation often neglects the potential for interaction between labeled and unlabeled images. In this paper, we introduce InterLUDE, a new approach to enhance SSL made of two parts that each benefit from labeled-unlabeled interaction. The first part, embedding fusion, interpolates between labeled and unlabeled embeddings to improve representation learning. The second part is a new loss, grounded in …

Go to article

Core-Periphery Multi-Modality Feature Alignment For Zero-Shot Medical Image Analysis, Xiaowei Yu, Lu Zhang, Zihao Wu, Dajiang Zhu Jan 2024

Core-Periphery Multi-Modality Feature Alignment For Zero-Shot Medical Image Analysis, Xiaowei Yu, Lu Zhang, Zihao Wu, Dajiang Zhu

Computer Science Faculty Research & Creative Works

Multi-modality learning, exemplified by the language-image pair pre-trained CLIP model, has demonstrated remarkable performance in enhancing zero-shot capabilities and has gained significant attention recently. However, simply applying language-image pre-trained CLIP to medical image analysis encounters substantial domain shifts, resulting in severe performance degradation due to inherent disparities between natural (non-medical) and medical image characteristics. To address this challenge and uphold or even enhance CLIP's zero-shot capability in medical image analysis, we develop a novel approach, Core-Periphery feature alignment for CLIP (CP-CLIP), to model medical images and corresponding clinical text jointly. To achieve this, we design an auxiliary neural network whose …

Go to article

Enhancing Group-Wise Consistency In 3-Hinge Gyrus Matching Via Anatomical Embedding And Structural Connectivity Optimization, Chao Cao, Xiaowei Yu, Lu Zhang, Tong Chen, Yanjun Lyu, Tianming Liu, Dajiang Zhu Jan 2024

Enhancing Group-Wise Consistency In 3-Hinge Gyrus Matching Via Anatomical Embedding And Structural Connectivity Optimization, Chao Cao, Xiaowei Yu, Lu Zhang, Tong Chen, Yanjun Lyu, Tianming Liu, Dajiang Zhu

Computer Science Faculty Research & Creative Works

Recently, a novel cortical folding pattern known as the 3-hinge gyrus (3HG) has been identified. 3HGs are defined as the convergence of the gyri coming from three distinct directions on gyral crests. In contrast to cortical regions, 3HGs are defined at a finer scale and they widely exist across different individuals, representing both commonalities and individualities of cortical folding patterns. It is important to note that 3HGs are identified in individual spaces, lacking natural cross-subject correspondences. To address this issue, we have developed a learning-based method to encode anatomical features of 3HGs into a set of embedding vectors that can …

Go to article

The Perils Of Wi-Fi Spoofing Attack Via Geolocation Api And Its Defense, Xiao Han, Junjie Xiong, Wenbo Shen, Mingkui Wei, Shangqing Zhao, Zhuo Lu, Yao Liu Jan 2024

The Perils Of Wi-Fi Spoofing Attack Via Geolocation Api And Its Defense, Xiao Han, Junjie Xiong, Wenbo Shen, Mingkui Wei, Shangqing Zhao, Zhuo Lu, Yao Liu

Computer Science Faculty Research & Creative Works

Location spoofing attack deceiving a Wi-Fi positioning system has been studied for over a decade. However, it has been challenging to construct a practical spoofing attack in urban areas with dense coverage of legitimate Wi-Fi APs. This paper identifies the vulnerability of the Google Geolocation API, which returns the location of a mobile device based on the information of the Wi-Fi access points that the device can detect. We show that this vulnerability can be exploited by the attacker to reveal the black-box localization algorithms adopted by the Google Wi-Fi positioning system and easily launch the location spoofing attack in …

Go to article

Toward The Unified Theory Of Said-Linked Subauroral Arcs, Evgeny V. Mishin, Anatoly V. Streltsov Jan 2024

Toward The Unified Theory Of Said-Linked Subauroral Arcs, Evgeny V. Mishin, Anatoly V. Streltsov

Publications

We present a unified approach to subauroral arcs within intense subauroral ion drifts (SAID), which explains the observed transition of a precursor Stable Auroral Red (SAR) arc into Strong Thermal Emission Velocity Enhancement (STEVE). This approach is based on the short-circuiting concept of fasttime SAID as an integral part of a magnetospheric voltage generator between the innermost boundaries of the freshly injected plasma sheet electrons and ring current ions. Here, enhanced plasma turbulence rapidly heats the bulk plasma and accelerates suprathermal non-Maxwellian “tails.” Heat and suprathermal electron transport rapidly elevate the ionospheric electron temperature—the source of a bright SAR arc. …

Go to article

Leveraging Ai And Big Data In Low-Resource Healthcare Settings, Ahmed Hussein Ali, Saad Ahmed Dheyab, Abdullah Hussein Alamoodi, Aws Abed Al Raheem Magableh, Yuantong Gu Jan 2024

Leveraging Ai And Big Data In Low-Resource Healthcare Settings, Ahmed Hussein Ali, Saad Ahmed Dheyab, Abdullah Hussein Alamoodi, Aws Abed Al Raheem Magableh, Yuantong Gu

Mesopotamian Journal of Big Data

Big data and artificial intelligence are game-changing technologies for the underdeveloped healthcare industry because they help optimize the entire supply chain and deliver more exact patient outcome information. Machine learning approaches that have recently seen more growing popularity include deep learning models that have brought revolution within the healthcare system in the previous years due to more complicated data compared to previous years . Machine learning is an essential data analysis procedure to describe efficient and effective methods to extract hidden information from large amounts of data that it would take logical analytics too long to manage. Recent years have …

Go to article

Enhancing Xml-Based Compiler Construction With Large Language Models: A Novel Approach, Idrees A. Zahid, Shahad Sabbar Joudar Jan 2024

Enhancing Xml-Based Compiler Construction With Large Language Models: A Novel Approach, Idrees A. Zahid, Shahad Sabbar Joudar

Mesopotamian Journal of Big Data

Considering the prevailing rule of Large Language Models (LLMs) applications and the benefits of XML in a compiler context. This manuscript explores the synergistic integration of Large Language Models with XML-based compiler tools and advanced computing technologies. Marking a significant stride toward redefining compiler construction and data representation paradigms. As computing power and internet proliferation advance, XML emerges as a pivotal technology for representing, exchanging, and transforming documents and data. This study builds on the foundational work of Chomsky's Context-Free Grammar (CFG). Recognized for their critical role in compiler construction, to address and mitigate the speed penalties associated with traditional …

Go to article

Mlp And Rbf Algorithms In Finance: Predicting And Classifying Stock Prices Amidst Economic Policy Uncertainty, Bushra Ali, Khder Alakkari, Mostafa Abotaleb, Maad M. Mijwil, Klodian Dhoska Jan 2024

Mlp And Rbf Algorithms In Finance: Predicting And Classifying Stock Prices Amidst Economic Policy Uncertainty, Bushra Ali, Khder Alakkari, Mostafa Abotaleb, Maad M. Mijwil, Klodian Dhoska

Mesopotamian Journal of Big Data

In the realm of stock market prediction and classification, the use of machine learning algorithms has gained significant attention. In this study, we explore the application of Multilayer Perceptron (MLP) and Radial Basis Function (RBF) algorithms in predicting and classifying stock prices, specifically amidst economic policy uncertainty. Stock market fluctuations are greatly influenced by economic policies implemented by governments and central banks. These policies can create uncertainty and volatility, which in turn makes accurate predictions and classifications of stock prices more challenging. By leveraging MLP and RBF algorithms, we aim to develop models that can effectively navigate these uncertainties and …

Go to article

Generalized Time Domain Prediction Model For Motor Imagery-Based Wheelchair Movement Control, Z. T. Al-Qaysi, M. S. Suzani, Nazre Bin Abdul Rashid, Reem D. Ismail, M. A. Ahmed, Rasha A. Aljanabi, Veronica Gil-Costa Jan 2024

Generalized Time Domain Prediction Model For Motor Imagery-Based Wheelchair Movement Control, Z. T. Al-Qaysi, M. S. Suzani, Nazre Bin Abdul Rashid, Reem D. Ismail, M. A. Ahmed, Rasha A. Aljanabi, Veronica Gil-Costa

Mesopotamian Journal of Big Data

Brain-computer interface (BCI-MI)-based wheelchair control is, in principle, an appropriate method for completely paralyzed people with a healthy brain. In a BCI-based wheelchair control system, pattern recognition in terms of preprocessing, feature extraction, and classification plays a significant role in avoiding recognition errors, which can lead to the initiation of the wrong command that will put the user in unsafe condition. Therefore, this research's goal is to create a time-domain generic pattern recognition model (GPRM) of two-class EEG-MI signals for use in a wheelchair control system. This GPRM has the advantage of having a model that is applicable to unknown …

Go to article

Hybrid Model For Forecasting Temperature In Khartoum Based On Cru Data, Hussein Alkattan, Alhumaima Ali Subhi, Laith Farhan, Ghazwan Al-Mashhadani Jan 2024

Hybrid Model For Forecasting Temperature In Khartoum Based On Cru Data, Hussein Alkattan, Alhumaima Ali Subhi, Laith Farhan, Ghazwan Al-Mashhadani

Mesopotamian Journal of Big Data

This consider leverages verifiable climatic data from the Climatic Research Unit (CRU), traversing from 1901 to 2022, to create progressed temperature forecasting models for Khartoum, Sudan. By applying state-of-the-art machine learning techniques, including Hybrid model, we aim to progress the precision of temperature forecasts in a semi-arid climate. The integration of long-term CRU data permits for the recognizable proof of climate patterns and patterns, upgrading the unwavering quality of short- and long-term forecasts. Moved forward temperature forecasting can altogether advantage basic segments empowering way better adjustment to climatic changes and extraordinary climate occasions. Our approach illustrates the potential of combining …

Go to article

Deep Learning Approaches For Gender Classification From Facial Images, Mustafa Abdulfattah Habeeb, Yahya Layth Khaleel, Reem D. Ismail, Z. T. Al-Qaysi, Fatimah N. Ameen Jan 2024

Deep Learning Approaches For Gender Classification From Facial Images, Mustafa Abdulfattah Habeeb, Yahya Layth Khaleel, Reem D. Ismail, Z. T. Al-Qaysi, Fatimah N. Ameen

Mesopotamian Journal of Big Data

Gender recognition on the facial level is considered one of the most important technologies that finds use in such fields as a personalized marketing plan, safe systems of authentication, and effective human-computer interfaces. However, it has the following challenges; variation of lighting, facial movement, and ethnic/age face images. AI and DL has been improving on the effectiveness, flexibility, and speed of the gender classification system. AI enables complex and automatic feature learning in Data, while DL is tailored for handle variants in vision-based data. In this paper, we evaluated several architectures including Efficient Net_B2, ResNet50, ResNet18, and Lightning whilst determining …

Go to article

Automated Water Quality Assessment Using Big Data Analytics, Yasmin Makki Mohialden, Nadia Mahmood Hussien, Saba Abdulbaqi Salman Jan 2024

Automated Water Quality Assessment Using Big Data Analytics, Yasmin Makki Mohialden, Nadia Mahmood Hussien, Saba Abdulbaqi Salman

Mesopotamian Journal of Big Data

Water is one of the world's most precious resources, essential to life. Industrial waste, agricultural runoff, and urban discharge degrade water, rendering it unfit for consumption. Water quality monitoring and evaluation are more important than ever. Big Data analytics is used to examine water quality utilizing enormous datasets of pH, hardness, solids concentration, chloramine, sulfate, conductivity, organic carbon, trihalomethanes, and turbidity. This work classifies water potability, which is vital for human consumption, using strong machine learning on massive datasets. Classifiers were Random Forest, Gradient Boosting, and Support Vector Machine on 3,276 water bodies. The Random Forest classifier obtained the highest …

Go to article

Anomaly-Based Ids (Intrusion Detection System) For Cyber-Physical Systems, Ahmad Muter Awaad, Khattab M. Alheeti, Abdul Kream A.H. Najem Jan 2024

Anomaly-Based Ids (Intrusion Detection System) For Cyber-Physical Systems, Ahmad Muter Awaad, Khattab M. Alheeti, Abdul Kream A.H. Najem

Mesopotamian Journal of Big Data

Cyber-physical systems (CPS) are critical infrastructures that integrate physical processes with computational components. The security of CPS is paramount, as any breach can lead to severe consequences. Anomaly-based intrusion detection systems (IDS) have emerged as a promising approach to safeguard CPS against cyber threats. This paper presents an anomaly-based IDS designed specifically for CPS, leveraging machine learning techniques to establish a baseline of normal system behaviour and promptly detect deviations indicative of malicious activities. The proposed system incorporates multiple classification techniques, including KNeighbors, RandomForest, XGB, DecisionTree, SGD, SVM, LGBM, AdaBoost, Bagging, and MLP Classifier, to enhance detection accuracy and robustness. …

Go to article

Catalog 2024-2025, University Of North Florida Jan 2024

Catalog 2024-2025, University Of North Florida

UNF Course Catalogs

No abstract provided.

Go to article

Soci+: An Enhanced Toolkit For Secure Outsourced Computation On Integers, Bowen Zhao, Weiquan Deng, Xiaoguo Li, Ximeng Liu, Qingqi Pei, Robert H. Deng Jan 2024

Soci+: An Enhanced Toolkit For Secure Outsourced Computation On Integers, Bowen Zhao, Weiquan Deng, Xiaoguo Li, Ximeng Liu, Qingqi Pei, Robert H. Deng

Research Collection School Of Computing and Information Systems

Secure outsourced computation is critical for cloud computing to safeguard data confidentiality and ensure data usability. Recently, secure outsourced computation schemes following a twin-server architecture based on partially homomorphic cryptosystems have received increasing attention. The Secure Outsourced Computation on Integers (SOCI) [1] toolkit is the state-of-the-art among these schemes which can perform secure computation on integers without requiring the costly bootstrapping operation as in fully homomorphic encryption; however, SOCI suffers from relatively large computation and communication overhead. In this paper, we propose SOCI+ which significantly improves the performance of SOCI. Specifically, SOCI+ employs a novel (2,2)-threshold Paillier cryptosystem with fast …

Go to article

Dynamic Meta-Path Guided Temporal Heterogeneous Graph Neural Networks, Yugang Ji, Chuan Shi, Yuan Fang Jan 2024

Dynamic Meta-Path Guided Temporal Heterogeneous Graph Neural Networks, Yugang Ji, Chuan Shi, Yuan Fang

Research Collection School Of Computing and Information Systems

Graph Neural Networks (GNNs) have become the de facto standard for representation learning on topological graphs, which usually derive effective node representations via message passing from neighborhoods. Although GNNs have achieved great success, previous models are mostly confined to static and homogeneous graphs. However, there are multiple dynamic interactions between different-typed nodes in real-world scenarios like academic networks and e-commerce platforms, forming temporal heterogeneous graphs (THGs). Limited work has been done for representation learning on THGs and the challenges are in two aspects. First, there are abundant dynamic semantics between nodes while traditional techniques like meta-paths can only capture static …

Go to article

A Unified Framework For Contextual And Factoid Question Generation, Chenhe Dong, Ying Shen, Shiyang Lin, Zhenzhou Lin, Yang Deng Jan 2024

A Unified Framework For Contextual And Factoid Question Generation, Chenhe Dong, Ying Shen, Shiyang Lin, Zhenzhou Lin, Yang Deng

Research Collection School Of Computing and Information Systems

Question generation (QG) aims to automatically generate fluent and relevant questions, where the two most mainstream directions are generating questions from unstructured contextual texts (CQG), such as news articles, and generating questions from structured factoid texts (FQG), such as knowledge graphs or tables. Existing methods for these two tasks mainly face challenges of limited internal structural information as well as scarce background information, while these two tasks can benefit each other for alleviating these issues. For example, when meeting the entity mention “United Kingdom” in CQG, it can be inferred that it is a country in European continent based on …

Go to article

Wakening Past Concepts Without Past Data: Class-Incremental Learning From Online Placebos, Yaoyao Liu, Yingying Li, Bernt Schiele, Qianru Sun Jan 2024

Wakening Past Concepts Without Past Data: Class-Incremental Learning From Online Placebos, Yaoyao Liu, Yingying Li, Bernt Schiele, Qianru Sun

Research Collection School Of Computing and Information Systems

Not forgetting old class knowledge is a key challenge for class-incremental learning (CIL) when the model continuously adapts to new classes. A common technique to address this is knowledge distillation (KD), which penalizes prediction inconsistencies between old and new models. Such prediction is made with almost new class data, as old class data is extremely scarce due to the strict memory limitation in CIL. In this paper, we take a deep dive into KD losses and find that "using new class data for KD"not only hinders the model adaption (for learning new classes) but also results in low efficiency for …

Go to article

Demonstrating Canvas-Based Processing Of Multiple Camera Streams At The Edge, Ila Gokarn, Hemanth Sabbella, Yigong Hu, Tarek Abdelzaher, Archan Misra Jan 2024

Demonstrating Canvas-Based Processing Of Multiple Camera Streams At The Edge, Ila Gokarn, Hemanth Sabbella, Yigong Hu, Tarek Abdelzaher, Archan Misra

Research Collection School Of Computing and Information Systems

We demonstrate criticality-aware canvas-based processing of multiple concurrent camera streams at the resource constrained edge to show substantial improvement in the accuracy-throughput trade-off. The proposed system focuses the available computation resources on select Regions of Interest (RoI) across all the camera streams by (i) extracting RoI from the input camera stream (ii) 2D bin packing the RoI on a canvas frame and (iii) batching and inferring upon these constructed composite canvas frames with a YOLOv5 object detection model. Our experiments show that such canvas-based processing can (i) sustain real-time processing throughput of 23 FPS per camera across 6 concurrent input …

Go to article

First Prev Next Last