Boosting Adversarial Training In Safety-Critical Systems Through Boundary Data Selection,
2023
Singapore Management University
Boosting Adversarial Training In Safety-Critical Systems Through Boundary Data Selection, Yifan Jia, Christopher M. Poskitt, Peixin Zhang, Jingyi Wang, Jun Sun, Sudipta Chattopadhyay
Research Collection School Of Computing and Information Systems
AI-enabled collaborative robots are designed to be used in close collaboration with humans, thus requiring stringent safety standards and quick response times. Adversarial attacks pose a significant threat to the deep learning models of these systems, making it crucial to develop methods to improve the models' robustness against them. Adversarial training is one approach to improve their robustness: it works by augmenting the training data with adversarial examples. This, unfortunately, comes with the cost of increased computational overhead and extended training times. In this work, we balance the need for additional adversarial data with the goal of minimizing the training …
Hrgcn: Heterogeneous Graph-Level Anomaly Detection With Hierarchical Relation-Augmented Graph Neural Networks,
2023
Singapore Management University
Hrgcn: Heterogeneous Graph-Level Anomaly Detection With Hierarchical Relation-Augmented Graph Neural Networks, Jiaxi Li, Guansong Pang, Ling Chen, Mohammad-Reza Namazi-Rad
Research Collection School Of Computing and Information Systems
This work considers the problem of heterogeneous graph-level anomaly detection. Heterogeneous graphs are commonly used to represent behaviours between different types of entities in complex industrial systems for capturing as much information about the system operations as possible. Detecting anomalous heterogeneous graphs from a large set of system behaviour graphs is crucial for many real-world applications like online web/mobile service and cloud access control. To address the problem, we propose HRGCN, an unsupervised deep heterogeneous graph neural network, to model complex heterogeneous relations between different entities in the system for effectively identifying these anomalous behaviour graphs. HRGCN trains a hierarchical …
Anomaly Detection Under Distribution Shift,
2023
Singapore Management University
Anomaly Detection Under Distribution Shift, Tri Cao, Jiawen Zhu, Guansong Pang
Research Collection School Of Computing and Information Systems
Anomaly detection (AD) is a crucial machine learning task that aims to learn patterns from a set of normal training samples to identify abnormal samples in test data. Most existing AD studies assume that the training and test data are drawn from the same data distribution, but the test data can have large distribution shifts arising in many real-world applications due to different natural variations such as new lighting conditions, object poses, or background appearances, rendering existing AD methods ineffective in such cases. In this paper, we consider the problem of anomaly detection under distribution shift and establish performance benchmarks …
Toward Intention Discovery For Early Malice Detection In Cryptocurrency,
2023
Singapore Management University
Toward Intention Discovery For Early Malice Detection In Cryptocurrency, Ling Cheng, Feida Zhu, Yong Wang, Ruicheng Liang, Huiwen Liu
Research Collection School Of Computing and Information Systems
Cryptocurrency’s pseudo-anonymous nature makes it vulnerable to malicious activities. However, existing deep learning solutions lack interpretability and only support retrospective analysis of specific malice types. To address these challenges, we propose Intention-Monitor for early malice detection in Bitcoin. Our model, utilizing Decision-Tree based feature Selection and Complement (DT-SC), builds different feature sets for different malice types. The Status Proposal Module (SPM) and hierarchical self-attention predictor provide real-time global status and address label predictions. A survival module determines the stopping point and proposes the status sequence (intention). Our model detects various malicious activities with strong interpretability, outperforming state-of-the-art methods in extensive …
Flacgec: A Chinese Grammatical Error Correction Dataset With Fine-Grained Linguistic Annotation,
2023
Singapore Management University
Flacgec: A Chinese Grammatical Error Correction Dataset With Fine-Grained Linguistic Annotation, Hanyue Du, Yike Zhao, Qingyuan Tian, Jiani Wang, Lei Wang, Yunshi Lan, Xuesong Lu
Research Collection School Of Computing and Information Systems
Chinese Grammatical Error Correction (CGEC) has been attracting growing attention from researchers recently. In spite of the fact that multiple CGEC datasets have been developed to support the research, these datasets lack the ability to provide a deep linguistic topology of grammar errors, which is critical for interpreting and diagnosing CGEC approaches. To address this limitation, we introduce FlaCGEC, which is a new CGEC dataset featured with fine-grained linguistic annotation. Specifically, we collect raw corpus from the linguistic schema defined by Chinese language experts, conduct edits on sentences via rules, and refine generated samples manually, which results in 10k sentences …
Invariant Training 2d-3d Joint Hard Samples For Few-Shot Point Cloud Recognition,
2023
Singapore Management University
Invariant Training 2d-3d Joint Hard Samples For Few-Shot Point Cloud Recognition, Xuanyu Yi, Jiajun Deng, Qianru Sun, Xian-Sheng Hua, Joo-Hwee Lim, Hanwang Zhang
Research Collection School Of Computing and Information Systems
We tackle the data scarcity challenge in few-shot point cloud recognition of 3D objects by using a joint prediction from a conventional 3D model and a well-pretrained 2D model. Surprisingly, such an ensemble, though seems trivial, has hardly been shown effective in recent 2D-3D models. We find out the crux is the less effective training for the “joint hard samples”, which have high confidence prediction on different wrong labels, implying that the 2D and 3D models do not collaborate well. To this end, our proposed invariant training strategy, called INVJOINT, does not only emphasize the training more on the hard …
Sentiment Analysis Of Public Perception Towards Elon Musk On Reddit (2008-2022),
2023
University of Louisville
Sentiment Analysis Of Public Perception Towards Elon Musk On Reddit (2008-2022), Daniel Maya Bonilla, Samuel Iradukunda, Pamela Thomas
The Cardinal Edge
As Elon Musk’s influence in technology and business continues to expand, it becomes crucial to comprehend public sentiment surrounding him in order to gauge the impact of his actions and statements. In this study, we conducted a comprehensive analysis of comments from various subreddits discussing Elon Musk over a 14-year period, from 2008 to 2022. Utilizing advanced sentiment analysis models and natural language processing techniques, we examined patterns and shifts in public sentiment towards Musk, identifying correlations with key events in his life and career. Our findings reveal that public sentiment is shaped by a multitude of factors, including his …
Blended Learning In The Wake Of Ict Infrastructure Deficiencies: The Case Of A Zimbabwean University,
2023
Women's University in Africa
Blended Learning In The Wake Of Ict Infrastructure Deficiencies: The Case Of A Zimbabwean University, Lucia Makwasha, Sam Jnr Takavarasha, Hazel Mubango
African Conference on Information Systems and Technology
In the wake of debates between actors in the Zimbabwean higher education sector about the effectiveness of e-learning models, it is important to investigate the effectiveness of using blended learning at a time when infrastructure challenges are disrupting ICT access. This paper aims to address this quest for a deeper understanding by investigating students' perceptions of blended learning at a selected Zimbabwean university. Twelve in-depth interviews were conducted with students from a Zimbabwean university that employs blended learning under an interpretivist paradigm. Vygotsky's Zone of Proximal Development (ZPD) was used for conceptualising students' cognitive development and Engestrom's (2003) Third-generation Activity …
When Routing Meets Recommendation: Solving Dynamic Order Recommendations Problem In Peer-To-Peer Logistics Platforms,
2023
Singapore Management University
When Routing Meets Recommendation: Solving Dynamic Order Recommendations Problem In Peer-To-Peer Logistics Platforms, Zhiqin Zhang, Waldy Joe, Yuyang Er, Hoong Chuin Lau
Research Collection School Of Computing and Information Systems
Peer-to-Peer (P2P) logistics platforms, unlike traditional last-mile logistics providers, do not have dedicated delivery resources (both vehicles and drivers). Thus, the efficiency of such operating model lies in the successful matching of demand and supply, i.e., how to match the delivery tasks with suitable drivers that will result in successful assignment and completion of the tasks. We consider a Same-Day Delivery Problem (SDDP) involving a P2P logistics platform where new orders arrive dynamically and the platform operator needs to generate a list of recommended orders to the crowdsourced drivers. We formulate this problem as a Dynamic Order Recommendations Problem (DORP). …
Rosas: Deep Semi-Supervised Anomaly Detection With Contamination-Resilient Continuous Supervision,
2023
Singapore Management University
Rosas: Deep Semi-Supervised Anomaly Detection With Contamination-Resilient Continuous Supervision, Hongzuo Xu, Yijie Wang, Guansong Pang, Songlei Jian, Ning Liu, Yongjun Wang
Research Collection School Of Computing and Information Systems
Semi-supervised anomaly detection methods leverage a few anomaly examples to yield drastically improved performance compared to unsupervised models. However, they still suffer from two limitations: 1) unlabeled anomalies (i.e., anomaly contamination) may mislead the learning process when all the unlabeled data are employed as inliers for model training; 2) only discrete supervision information (such as binary or ordinal data labels) is exploited, which leads to suboptimal learning of anomaly scores that essentially take on a continuous distribution. Therefore, this paper proposes a novel semi-supervised anomaly detection method, which devises contamination-resilient continuous supervisory signals. Specifically, we propose a mass interpolation method …
Generative Model-Based Testing On Decision-Making Policies,
2023
Kyushu University
Generative Model-Based Testing On Decision-Making Policies, Zhuo Li, Xiongfei Wu, Derui Zhu, Mingfei Cheng, Siyuan Chen, Fuyuan Zhang, Xiaofei Xie, Lei Ma, Jianjun Zhao
Research Collection School Of Computing and Information Systems
The reliability of decision-making policies is urgently important today as they have established the fundamentals of many critical applications, such as autonomous driving and robotics. To ensure reliability, there have been a number of research efforts on testing decision-making policies that solve Markov decision processes (MDPs). However, due to the deep neural network (DNN)-based inherit and infinite state space, developing scalable and effective testing frameworks for decision-making policies still remains open and challenging.In this paper, we present an effective testing framework for decision-making policies. The framework adopts a generative diffusion model-based test case generator that can easily adapt to different …
Graph-Level Anomaly Detection Via Hierarchical Memory Networks,
2023
Singapore Management University
Graph-Level Anomaly Detection Via Hierarchical Memory Networks, Chaoxi Niu, Guansong Pang, Ling Chen
Research Collection School Of Computing and Information Systems
Graph-level anomaly detection aims to identify abnormal graphs that exhibit deviant structures and node attributes compared to the majority in a graph set. One primary challenge is to learn normal patterns manifested in both fine-grained and holistic views of graphs for identifying graphs that are abnormal in part or in whole. To tackle this challenge, we propose a novel approach called Hierarchical Memory Networks (HimNet), which learns hierarchical memory modules---node and graph memory modules---via a graph autoencoder network architecture. The node-level memory module is trained to model fine-grained, internal graph interactions among nodes for detecting locally abnormal graphs, while the …
Understanding Multi-Homing And Switching By Platform Drivers,
2023
Singapore Management University
Understanding Multi-Homing And Switching By Platform Drivers, Xiaotong Guo, Andreas Haupt, Hai Wang, Rida Qadri, Jinhua Zhao
Research Collection School Of Computing and Information Systems
Freelance drivers in the shared mobility market frequently switch or work for multiple platforms, affecting driver labor supply. Due to the importance of driver labor supply for the shared mobility market, understanding drivers’ switching and multi-homing behavior is vital to managing service quality on – and effective regulation of – mobility platforms. However, a lack of individual-level data on driver behavior has thus far impeded a deeper understanding. This paper taxonomizes and estimates perceived switching and multi-homing frictions on mobility platforms. Based on a structural model of driver labor supply, we estimate switching and multi-homing costs in a platform duopoly …
The Devil Is In The Tails: How Long-Tailed Code Distributions Impact Large Language Models,
2023
Singapore Management University
The Devil Is In The Tails: How Long-Tailed Code Distributions Impact Large Language Models, Xin Zhou, Kisub Kim, Bowen Xu, Jiakun Liu, Donggyun Han, David Lo
Research Collection School Of Computing and Information Systems
Learning-based techniques, especially advanced Large Language Models (LLMs) for code, have gained considerable popularity in various software engineering (SE) tasks. However, most existing works focus on designing better learning-based models and pay less attention to the properties of datasets. Learning-based models, including popular LLMs for code, heavily rely on data, and the data's properties (e.g., data distribution) could significantly affect their behavior. We conducted an exploratory study on the distribution of SE data and found that such data usually follows a skewed distribution (i.e., long-tailed distribution) where a small number of classes have an extensive collection of samples, while a …
Real: A Representative Error-Driven Approach For Active Learning,
2023
Singapore Management University
Real: A Representative Error-Driven Approach For Active Learning, Cheng Chen, Yong Wang, Lizi Liao, Yueguo Chen, Xiaoyong Du
Research Collection School Of Computing and Information Systems
Given a limited labeling budget, active learning (al) aims to sample the most informative instances from an unlabeled pool to acquire labels for subsequent model training. To achieve this, al typically measures the informativeness of unlabeled instances based on uncertainty and diversity. However, it does not consider erroneous instances with their neighborhood error density, which have great potential to improve the model performance. To address this limitation, we propose Real, a novel approach to select data instances with Representative Errors for Active Learning. It identifies minority predictions as pseudo errors within a cluster and allocates an adaptive sampling budget for …
Threshold Attribute-Based Credentials With Redactable Signature,
2023
Singapore Management University
Threshold Attribute-Based Credentials With Redactable Signature, Rui Shi, Huamin Feng, Yang Yang, Feng Yuan, Yingjiu Li, Hwee Hwa Pang, Robert H. Deng
Research Collection School Of Computing and Information Systems
Threshold attribute-based credentials are suitable for decentralized systems such as blockchains as such systems generally assume that authenticity, confidentiality, and availability can still be guaranteed in the presence of a threshold number of dishonest or faulty nodes. Coconut (NDSS'19) was the first selective disclosure attribute-based credentials scheme supporting threshold issuance. However, it does not support threshold tracing of user identities and threshold revocation of user credentials, which is desired for internal governance such as identity management, data auditing, and accountability. The communication and computation complexities of Coconut for verifying credentials are linear in the number of each user's attributes and …
Continual Collaborative Filtering Through Gradient Alignment,
2023
Singapore Management University
Continual Collaborative Filtering Through Gradient Alignment, Dinh Hieu Do, Hady Wirawan Lauw
Research Collection School Of Computing and Information Systems
A recommender system operates in a dynamic environment where new items emerge and new users join the system, resulting in ever-growing user-item interactions over time. Existing works either assume a model trained offline on a static dataset (requiring periodic re-training with ever larger datasets); or an online learning setup that favors recency over history. As privacy-aware users could hide their histories, the loss of older information means that periodic retraining may not always be feasible, while online learning may lose sight of users' long-term preferences. In this work, we adopt a continual learning perspective to collaborative filtering, by compartmentalizing users …
On Predicting Esg Ratings Using Dynamic Company Networks,
2023
Singapore Management University
On Predicting Esg Ratings Using Dynamic Company Networks, Gary Ang, Zhiling Guo, Ee-Peng Lim
Research Collection School Of Computing and Information Systems
Environmental, social and governance (ESG) considerations play an increasingly important role due to the growing focus on sustainability globally. Entities, such as banks and investors, utilize ESG ratings of companies issued by specialized rating agencies to evaluate ESG risks of companies. The process of assigning ESG ratings by human analysts is however laborious and time intensive. Developing methods to predict ESG ratings could alleviate such challenges, allow ESG ratings to be generated in a more timely manner, cover more companies, and be more accessible. Most works study the effects of ESG ratings on target variables such as stock prices or …
Models And Algorithms For Promoting Diverse And Fair Query Results,
2023
New Jersey Institute of Technology
Models And Algorithms For Promoting Diverse And Fair Query Results, Md Mouinul Islam
Dissertations
Ensuring fairness and diversity in search results are two key concerns in compelling search and recommendation applications. This work explicitly studies these two aspects given multiple users' preferences as inputs, in an effort to create a single ranking or top-k result set that satisfies different fairness and diversity criteria. From group fairness standpoint, it adapts demographic parity like group fairness criteria and proposes new models that are suitable for ranking or producing top-k set of results. This dissertation also studies equitable exposure of individual search results in long tail data, a concept related to individual fairness. First, the dissertation focuses …
Diversification And Fairness In Top-K Ranking Algorithms,
2023
New Jersey Institute of Technology
Diversification And Fairness In Top-K Ranking Algorithms, Mahsa Asadi
Dissertations
Given a user query, the typical user interfaces, such as search engines and recommender systems, only allow a small number of results to be returned to the user. Hence, figuring out what would be the top-k results is an important task in information retrieval, as it helps to ensure that the most relevant results are presented to the user. There exists an extensive body of research that studies how to score the records and return top-k to the user. Moreover, there exists an extensive set of criteria that researchers identify to present the user with top-k results, and result diversification …
