Computer Sciences Commons^™

Asynchronous Messaging In A P2p System: Defending Against A Storage Exhaustion Attack On Kademlia Dht, Maxim Biro

Theses and Dissertations

An instant messaging service designed using a peer to peer distributed network architecture has many appealing properties it gets for free: high scalability, cheap operational cost and no reliance on a third party to provide the service. However, the nature of the distributed network architecture makes implementing some of the instant messaging features rather challenging, asynchronous messaging being one of them. The asynchronous messaging requires that peers store arbitrary data on behalf of other peers for prolonged periods of time, often measured in days, which, if not kept in check, can be easily abused by malicious actors by spamming the …

Process For Designing And Implementing Provably Verifiable Voting Systems, Kholud Alghamdi

Theses and Dissertations

This research aims to explore processes for designing verifiable voting systems in which certain properties can be proven, exemplified with systems applicable to election processes in Saudi Arabia. The electronic government model has become a substantial channel for governments to connect to businesses and citizens, to develop services, and provide general information. E-voting and in particular online voting is one of the important tools that can be used in political and administrative places where information and communications technology devices and tools are utilized to simplify people’s lives and facilitate the election process and decision making. Election processes let a population …

Learning Depth From Images, Zhenyao Wu

Theses and Dissertations

Estimating depth from images has become a very popular task in computer vision which aims to restore the 3D scene from 2D images and identify important geometric knowledge of the scene. Its performance has been significantly improved by convolutional neural networks in recent years, which surpass the traditional methods by a large margin. However, the natural scenes are usually complicated, and hard to build the correspondence between pixels across frames, such as the region containing moving objects, illumination changes, occlusions, and reflections. This research explores rich and comprehensive spatial correspondence across images and designs three new network architectures for depth …

Strategic Signaling For Utility Control In Audit Games, Jianan Chen, Qin Hu, Honglu Jiang

Informatics and Engineering Systems Faculty Publications

As an effective method to protect the daily access to sensitive data against malicious attacks, the audit mechanism has been widely deployed in various practical fields. In order to examine security vulnerabilities and prevent the leakage of sensitive data in a timely manner, the database logging system usually employs an online signaling scheme to issue an alert when suspicious access is detected. Defenders can audit alerts to reduce potential damage. This interaction process between a defender and an attacker can be modeled as an audit game. In previous studies, it was found that sending real-time signals in the audit …

Deep Convolution Neural Networks For Image Classification, Arun D. Kulkarni

Computer Science Faculty Publications and Presentations

Deep learning is a highly active area of research in machine learning community. Deep Convolutional Neural Networks (DCNNs) present a machine learning tool that enables the computer to learn from image samples and extract internal representations or properties underlying grouping or categories of the images. DCNNs have been used successfully for image classification, object recognition, image segmentation, and image retrieval tasks. DCNN models such as Alex Net, VGG Net, and Google Net have been used to classify large dataset having millions of images into thousand classes. In this paper, we present a brief review of DCNNs and results of our …

Self-Guided Learning To Denoise For Robust Recommendation, Yunjun Gao, Yuntao Du, Yujia Hu, Lu Chen, Xinjun Zhu, Ziquan Fang, Baihua Zheng

Research Collection School Of Computing and Information Systems

The ubiquity of implicit feedback makes them the default choice to build modern recommender systems. Generally speaking, observed interactions are considered as positive samples, while unobserved interactions are considered as negative ones. However, implicit feedback is inherently noisy because of the ubiquitous presence of noisy-positive and noisy-negative interactions. Recently, some studies have noticed the importance of denoising implicit feedback for recommendations, and enhanced the robustness of recommendation models to some extent. Nonetheless, they typically fail to (1) capture the hard yet clean interactions for learning comprehensive user preference, and (2) provide a universal denoising solution that can be applied to …

Encoding Color Sequences In Active Tile Self-Assembly, Sonya Cirlos

Theses and Dissertations

Constructing patterns is a well-studied problem in both theoretical and experimental self-assembly with much of the work focused on multi-staged assembly. In this paper, we study building 1D patterns in a model of active self assembly: Tile Automata. This is a generalization of the 2-handed assembly model that borrows the concept of state changes from Cellular Automata. In this work we further develop the model by partitioning states as colors and show lower and upper bounds for building patterned assemblies based on an input pattern. Our first two sections utilize recent results to build binary strings along …

Predicting The Impact Of Iot Data Gathering On User’S Privacy Preferences, Ghassen Kilani

Theses and Dissertations

The proliferation of Internet of Things (IoT) devices has increased data sharing, profiling, and manipulation on various networks. The rapid growth of information disclosure has caused system users to lose motivation to enhance their data privacy. The repeated breaches on different networks worldwide have made people feel discouraged, as they perceive privacy schemes as futile. IoT systems introduce another dimension of privacy leakage due to their expendability nature and information collection features. The situation worsens when users have to manage multiple IoT devices, each following different security protocols, leading to poor decision-making and privacy leakage. This tremendous flow of unsecured …

Lead1.0: A Large-Scale Annotated Dataset For Energy Anomaly Detection In Commercial Buildings, Manoj Gulati, Pandarasamy Arjunan

Research Collection School Of Computing and Information Systems

Modern buildings are densely equipped with smart energy meters, which periodically generate a massive amount of time-series data yielding a few million data points every day. This data can be leveraged to discover the underlying load and infer their energy consumption patterns, inter-dependencies on environmental factors, and the building’s operational properties. Furthermore, it allows us to simultaneously identify anomalies present in the electricity consumption profiles, which is a big step towards saving energy and achieving global sustainability. However, to date, the lack of large-scale annotated energy consumption datasets hinders the ongoing research in anomaly detection. We contribute to this effort …

A3gan: Attribute-Aware Anonymization Networks For Face De-Identification, Liming Zhai, Qing Guo, Xiaofei Xie, Lei Ma, Yi Estelle Wang, Yang Liu

Research Collection School Of Computing and Information Systems

Face de-identification (De-ID) removes face identity information in face images to avoid personal privacy leakage. Existing face De-ID breaks the raw identity by cutting out the face regions and recovering the corrupted regions via deep generators, which inevitably affect the generation quality and cannot control generation results according to subsequent intelligent tasks (e.g., facial expression recognition). In this work, for the first attempt, we think the face De-ID from the perspective of attribute editing and propose an attribute-aware anonymization network (A3GAN) by formulating face De-ID as a joint task of semantic suppression and controllable attribute injection. Intuitively, the semantic suppression …

Towards Aligning Slides And Video Snippets: Mitigating Sequence And Content Mismatches, Ziyuan Liu, Hady W. Lauw

Research Collection School Of Computing and Information Systems

Slides are important form of teaching materials used in various courses at academic institutions. Due to their compactness, slides on their own may not stand as complete reference materials. To aid students’ understanding, it would be useful to supplement slides with other materials such as online videos. Given a deck of slides and a related video, we seek to align each slide in the deck to a relevant video snippet, if any. While this problem could be formulated as aligning two time series (each involving a sequence of text contents), we anticipate challenges in generating matches arising from differences in …

Declaration-Based Prompt Tuning For Visual Question Answering, Yuhang Liu, Wei Wei, Feida Zhu, Feida Zhu

Research Collection School Of Computing and Information Systems

In recent years, the pre-training-then-fine-tuning paradigm has yielded immense success on a wide spectrum of cross-modal tasks, such as visual question answering (VQA), in which a visual-language (VL) model is first optimized via self-supervised task objectives, e.g., masked language modeling (MLM) and image-text matching (ITM), and then fine-tuned to adapt to downstream task (e.g., VQA) via a brand-new objective function, e.g., answer prediction. However, the inconsistency of the objective forms not only severely limits the generalization of pre-trained VL models to downstream tasks, but also requires a large amount of labeled data for fine-tuning. To alleviate the problem, we propose …

Multi-Level Cross-View Contrastive Learning For Knowledge-Aware Recommender System, Ding Zou, Wei Wei, Xian-Ling Mao, Ziyang Wang, Minghui Qiu, Feida Zhu, Xin Cao

Research Collection School Of Computing and Information Systems

Knowledge graph (KG) plays an increasingly important role in recommender systems. Recently, graph neural networks (GNNs) based model has gradually become the theme of knowledge-aware recommendation (KGR). However, there is a natural deficiency for GNN-based KGR models, that is, the sparse supervised signal problem, which may make their actual performance drop to some extent. Inspired by the recent success of contrastive learning in mining supervised signals from data itself, in this paper, we focus on exploring the contrastive learning in KG-aware recommendation and propose a novel multi-level cross-view contrastive learning mechanism, named MCCLK. Different from traditional contrastive learning methods which …

Uncovering Inclusivity Gaps In Design Pedagogy Through The Digital Design Marginalization Framework, Jaisie Sin, Cosmin Munteanu, Michael Nixon, Velian Pandeliev, Garreth W. Tigwell, Kristen Shinohara, Anthony Tang, Steve Szigeti

Research Collection School Of Computing and Information Systems

Designers play a key role in the design of inclusive and socially conscious interfaces. Thus, it is imperative for designers to be thoughtful of the ethical and social implications of design. However, gaps in the foundational training that designers receive (e.g., as university students) can negatively impact their ability to consider the social implications of their design practice. This can result in consequences such as digital marginalization, which, as defined by the Digital Design Marginalization (DDM) framework, is the “pushing away”, whether intentional or not, of a defined group of users from a digital or online service or system, where …

An Empirical Study On Data Distribution-Aware Test Selection For Deep Learning Enhancement, Qiang Hu, Yuejun Guo, Maxime Cordy, Xiaofei Xie, Lei Ma, Mike Papadakis, Yves Le Traon

Research Collection School Of Computing and Information Systems

Similar to traditional software that is constantly under evolution, deep neural networks need to evolve upon the rapid growth of test data for continuous enhancement (e.g., adapting to distribution shift in a new environment for deployment). However, it is labor intensive to manually label all of the collected test data. Test selection solves this problem by strategically choosing a small set to label. Via retraining with the selected set, deep neural networks will achieve competitive accuracy. Unfortunately, existing selection metrics involve three main limitations: (1) using different retraining processes, (2) ignoring data distribution shifts, and (3) being insufficiently evaluated. To …

Finding Top-M Leading Records In Temporal Data, Yiyi Wang

Dissertations and Theses Collection (Open Access)

A traditional top-k query retrieves the records that stand out at a certain point in time. On the other hand, a durable top-k query considers how long the records retain their supremacy, i.e., it reports those records that are consistently among the top-k in a given time interval. In this thesis, we introduce a new query to the family of durable top-k formulations. It finds the top-m leading records, i.e., those that rank among the top-k for the longest duration within the query interval. Practically, this query assesses the records based on how long …

Towards Improving System Performance In Large Scale Multi-Agent Systems With Selfish Agents, Rajiv Ranjan Kumar

Dissertations and Theses Collection (Open Access)

Intelligent agents are becoming increasingly prevalent in a wide variety of domains including but not limited to transportation, safety and security. To better utilize the intelligence, there has been increasing focus on frameworks and methods for coordinating these intelligent agents. This thesis is specifically targeted at providing solution approaches for improving large scale multi-agent systems with selfish intelligent agents. In such systems, the performance of an agent depends on not just his/her own efforts, but also on other agent’s decisions. The complexity of interactions among multiple agents, coupled with the large scale nature of the problem domains and the uncertainties …

Deepcause: Verifying Neural Networks With Abstraction Refinement, Nguyen Hua Gia Phuc

Dissertations and Theses Collection (Open Access)

Neural networks have been becoming essential parts in many safety-critical systems (such
as self-driving cars and medical diagnosis). Due to that, it is desirable that neural networks
not only have high accuracy (which traditionally can be validated using a test set) but also
satisfy some safety properties (such as robustness, fairness, or free of backdoor). To verify
neural networks against desired safety properties, there are many approaches developed
based on classical abstract interpretation. However, like in program verification, these
approaches suffer from false alarms, which may hinder the deployment of the networks.

One natural remedy to tackle the problem adopted …

A Machine Learning Approach To Forecasting Sep Intensity And Times Based On Cme And Other Solar Activities, Peter John Thomas

Theses and Dissertations

High intensity Solar Energetic Particle (SEP) events pose severe risks for astronauts and critical infrastructure. The ability to accurately forecast the peak intensity and times of these events would enable preparatory measures to mitigate much of this risk. Machine learning approaches have the potential to use characteristics of CMEs and other space weather phenomena to predict SEP intensities and times. However, the severe sparsity of SEP events in current datasets poses a problem to traditional machine learning techniques. In this work, we present a dataset of proton event intensities and times, as well as features for corresponding CMEs and space …

An Ontological Approach For Recommending A Feature Selection Algorithm, Aparna Nayak, Bojan Bozic, Luca Longo

Conference papers

Feature selection plays an important role in machine learning or data mining problems. Removing irrelevant features increases model accuracy and reduces the computational cost. However, selecting important features is not a simple task as one feature selection algorithm does not perform well on all the datasets that are of interest. This paper tries to address the recommendation of a feature selection algorithm based on dataset characteristics and quality. The research uses three types of dataset characteristics along with data quality metrics. The main contribution of the work is the utilization of Semantic Web techniques to develop a novel system that …

Techniques To Enhance A Qubo Solver For Permutation-Based Combinatorial Optimization, Siong Thye Goh, Jianyuan Bo, Sabrish Gopalakrishnan, Hoong Chuin Lau

Research Collection School Of Computing and Information Systems

Many combinatorial optimization problems can be formulated as a problem to determine the order of sequence or to find a corresponding mapping of the objects. We call such problems permutation-based optimization problems. Many such problems can be formulated as a quadratic unconstrained binary optimization (QUBO) or Ising model by introducing a penalty coefficient to the permutation constraint terms. While classical and quantum annealing approaches have been proposed to solve QUBOs to date, they face issues with optimality and feasibility. Here we treat a given QUBO solver as a black box and propose techniques to enhance its performance. First, to ensure …

Contours Of Virtual Enfreakment In Fighting Game Characters, Sercan Sengun, Peter Mawhorter, James Bowie-Wilson, Yusef Audeh, Haewoon Kwak, D. Fox Harrell

Research Collection School Of Computing and Information Systems

Characters in fighting videogames1 such as Street Fighter V and Tekken7 typically reveal a phenomenon that we define as virtual enfreakment: their bodies, costumes, and fighting styles are exaggerated (1) in a manner that emphasizes perceived exoticism and (2) to enable them to be easily visually and conceptually distinguishable from one another. Here, using both quantitative and qualitative methods, including crowd-sourced surveys and analyses of game mechanics, we report on the contours of virtual enfreakment in those games. We specifically examine differences in character design across gender, national-origin, and skin-color lines. Disappointingly but not surprisingly, we find racism and sexism …

Multi-Agent Reinforcement Learning For Traffic Signal Control Through Universal Communication Method, Qize Jiang, Minhao Qin, Shengmin Shi, Weiwei Sun Sun, Baihua Zheng

Research Collection School Of Computing and Information Systems

How to coordinate the communication among intersections effectively in real complex traffic scenarios with multi-intersection is challenging. Existing approaches only enable the communication in a heuristic manner without considering the content/importance of information to be shared. In this paper, we propose a universal communication form UniComm between intersections. UniComm embeds massive observations collected at one agent into crucial predictions of their impact on its neighbors, which improves the communication efficiency and is universal across existing methods. We also propose a concise network UniLight to make full use of communications enabled by UniComm. Experimental results on real datasets demonstrate that UniComm …

Ai-Enabled Adaptive Learning Using Automated Topic Alignment And Doubt Detection, Kar Way Tan, Siaw Ling Lo, Eng Lieh Ouh, Wei Leng Neo

Research Collection School Of Computing and Information Systems

Implementing adaptive learning is often a challenging task at higher learning institutions where the students come from diverse backgrounds and disciplines. In this work, we collected informal learning journals from learners. Using the journals, we trained two machine learning models, an automated topic alignment and a doubt detection model to identify areas of adjustment required for teaching and students who require additional attention. The models form the baseline for a quiz recommender tool to dynamically generate personalized quizzes for each learner as practices to reinforce learning. Our pilot deployment of our AI-enabled Adaptive Learning System showed that our approach delivers …

Data-Driven Retail Decision-Making Using Spatial Partitioning And Delineation Of Communities, Ming Hui Tan, Kar Way Tan

Research Collection School Of Computing and Information Systems

Urbanisation is resulting in rapid growth in road networks within cities. The evolution of road networks can be indicative of a city's economic growth and it is a field of research gaining prominence in recent years. This paper proposes a framework for spatial partition of large scale road networks that produces appropriately sized geospatial units in order to identify the type of community they serve. To this end, we have developed a three-stage procedure which first partitions the road network using Louvain method, followed by outlining the boundary of each partition using Uber H3 grids before classifying each partition using …

Review Of Some Existing Qml Frameworks And Novel Hybrid Classical-Quantum Neural Networks Realising Binary Classification For The Noisy Datasets, N. Schetakis, D. Aghamalyan, Paul Robert Griffin, M. Boguslavsky

Research Collection School Of Computing and Information Systems

One of the most promising areas of research to obtain practical advantage is Quantum Machine Learning which was born as a result of cross-fertilisation of ideas between Quantum Computing and Classical Machine Learning. In this paper, we apply Quantum Machine Learning (QML) frameworks to improve binary classification models for noisy datasets which are prevalent in financial datasets. The metric we use for assessing the performance of our quantum classifiers is the area under the receiver operating characteristic curve AUC-ROC. By combining such approaches as hybrid-neural networks, parametric circuits, and data re-uploading we create QML inspired architectures and utilise them for …

Docee: A Large-Scale And Fine-Grained Benchmark For Document-Level Event Extraction, Meihan Tong, Bin Xu, Shuai Wang, Meihuan Han, Yixin Cao, Jiangqi Zhu, Siyu Chen, Lei Hou, Juanzi Li

Research Collection School Of Computing and Information Systems

Event extraction aims to identify an event and then extract the arguments participating in the event. Despite the great success in sentencelevel event extraction, events are more naturally presented in the form of documents, with event arguments scattered in multiple sentences. However, a major barrier to promote documentlevel event extraction has been the lack of large-scale and practical training and evaluation datasets. In this paper, we present DocEE, a new document-level event extraction dataset including 27,000+ events, 180,000+ arguments. We highlight three features: largescale manual annotations, fine-grained argument types and application-oriented settings. Experiments show that there is still a big …

End-To-End Open-Set Semi-Supervised Node Classification With Out-Of-Distribution Detection, Tiancheng Huang, Donglin Wang, Yuan Fang

Research Collection School Of Computing and Information Systems

Out-Of-Distribution (OOD) samples are prevalent in real-world applications. The OOD issue becomes even more severe on graph data, as the effect of OOD nodes can be potentially amplified by propagation through the graph topology. Recent works have considered the OOD detection problem, which is critical for reducing the uncertainty in learning and improving the robustness. However, no prior work considers simultaneously OOD detection and node classification on graphs in an end-to-end manner. In this paper, we study a novel problem of end-to-end open-set semisupervised node classification (OSSNC) on graphs, which deals with node classification in the presence of OOD nodes. …

Learning To Ask Critical Questions For Assisting Product Search, Zixuan Li, Lizi Liao, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Product search plays an essential role in eCommerce. It was treated as a special type of information retrieval problem. Most existing works make use of historical data to improve the search performance, which do not take the opportunity to ask for user’s current interest directly. Some session-aware methods take the user’s clicks within the session as implicit feedback, but it is still just a guess on user’s preference. To address this problem, recent conversational or question-based search models interact with users directly for understanding the user’s interest explicitly. However, most users do not have a clear picture on what to …

A Weakly Supervised Propagation Model For Rumor Verification And Stance Detection With Multiple Instance Learning, Ruichao Yang, Jing Ma, Hongzhan Lin, Wei Gao

Research Collection School Of Computing and Information Systems

The diffusion of rumors on social media generally follows a propagation tree structure, which provides valuable clues on how an original message is transmitted and responded by users over time. Recent studies reveal that rumor verification and stance detection are two relevant tasks that can jointly enhance each other despite their differences. For example, rumors can be debunked by cross-checking the stances conveyed by their relevant posts, and stances are also conditioned on the nature of the rumor. However, stance detection typically requires a large training set of labeled stances at post level, which are rare and costly to annotate. …