Computer Sciences Commons^™

Mosaic: Spatially-Multiplexed Edge Ai Optimization Over Multiple Concurrent Video Sensing Streams, Ila Gokarn, Hemanth Sabbella, Yigong Hu, Tarek Abdelzaher, Archan Misra

Research Collection School Of Computing and Information Systems

Sustaining high fidelity and high throughput of perception tasks over vision sensor streams on edge devices remains a formidable challenge, especially given the continuing increase in image sizes (e.g., generated by 4K cameras) and complexity of DNN models. One promising approach involves criticality-aware processing, where the computation is directed selectively to "critical" portions of individual image frames. We introduce MOSAIC, a novel system for such criticality-aware concurrent processing of multiple vision sensing streams that provides a multiplicative increase in the achievable throughput with negligible loss in perception fidelity. MOSAIC determines critical regions from images received from multiple vision …

A Mixed-Integer Linear Programming Reduction Of Disjoint Bilinear Programs Via Symbolic Variable Elimination, Jihwan Jeong, Scott Sanner, Akshat Kumar

Research Collection School Of Computing and Information Systems

A disjointly constrained bilinear program (DBLP) has various practical and industrial applications, e.g., in game theory, facility location, supply chain management, and multi-agent planning problems. Although earlier work has noted the equivalence of DBLP and mixed-integer linear programming (MILP) from an abstract theoretical perspective, a practical and exact closed-form reduction of a DBLP to a MILP has remained elusive. Such explicit reduction would allow us to leverage modern MILP solvers and techniques along with their solution optimality and anytime approximation guarantees. To this end, we provide the first constructive closed-form MILP reduction of a DBLP by extending the technique of …

Law Smells: Defining And Detecting Problematic Patterns In Legal Drafting, Corinna Coupette, Dirk Hartung, Janis Beckedorf, Maximilian Böther, Daniel Martin Katz

Research Collection Yong Pung How School Of Law

Building on the computer science concept of code smells, we initiate the study of law smells, i.e., patterns in legal texts that pose threats to the comprehensibility and maintainability of the law. With five intuitive law smells as running examples—namely, duplicated phrase, long element, large reference tree, ambiguous syntax, and natural language obsession—, we develop a comprehensive law smell taxonomy. This taxonomy classifies law smells by when they can be detected, which aspects of law they relate to, and how they can be discovered. We introduce textbased and graph-based methods to identify instances of law smells, confirming their utility in …

Law Smells: Defining And Detecting Problematic Patterns In Legal Drafting, Corinna Coupette, Dirk Hartung, Janis Beckedorf, Maximilian Bother, Daniel Martin Katz

Research Collection Yong Pung How School Of Law

Building on the computer science concept of code smells, we initiate the study of law smells, i.e., patterns in legal texts that pose threats to the comprehensibility and maintainability of the law. With five intuitive law smells as running examples—namely, duplicated phrase, long element, large reference tree, ambiguous syntax, and natural language obsession—, we develop a comprehensive law smell taxonomy. This taxonomy classifies law smells by when they can be detected, which aspects of law they relate to, and how they can be discovered. We introduce text-based and graph-based methods to identify instances of law smells, confirming their utility in …

Ocapo: Occupancy-Aware, Pdc Control For Open-Plan, Shared Workspaces, Anaradha Ravi, Archan Misra

Research Collection School Of Computing and Information Systems

Passive Displacement Cooling (PDC) has gained popularity as a means of significantly reducing building energy consumption overheads, especially in tropical climates. PDC eliminates the use of mechanical fans, instead using chilled-water heat exchangers to perform convective cooling. In this paper, we evaluate the impact of different parameters affecting occupant comfort in a 1000m2 open-floor area (consisting of multiple zones) of a ZEB (Zero Energy Building) deployed with PDC units and tackle the problem of setting the temperature setpoint of the PDC units to assure occupant thermal comfort. We tackle two key practical challenges: (a) the zone-level (i.e., occupant-experienced) temperature differs …

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

Research Collection School Of Computing and Information Systems

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

Groundnlq @ Ego4d Natural Language Queries Challenge 2023, Zhijian Hou, Lei Ji, Difei Gao, Wanjun Zhong, Kun Yan, Chong-Wah Ngo, Wing-Kwong Chan, Chong-Wah Ngo, Nan Duan, Mike Zheng Shou

Research Collection School Of Computing and Information Systems

In this report, we present our champion solution for Ego4D Natural Language Queries (NLQ) Challenge in CVPR 2023. Essentially, to accurately ground in a video, an effective egocentric feature extractor and a powerful grounding model are required. Motivated by this, we leverage a two-stage pre-training strategy to train egocentric feature extractors and the grounding model on video narrations, and further fine-tune the model on annotated data. In addition, we introduce a novel grounding model GroundNLQ, which employs a multi-modal multiscale grounding module for effective video and text fusion and various temporal intervals, especially for long videos. On the blind test …

Distpreserv: Maintaining User Distribution For Privacy-Preserving Location-Based Services, Yanbing Ren, Xinghua Li, Yinbin Miao, Robert H. Deng, Jian Weng, Siqi Ma, Jianfeng Ma

Research Collection School Of Computing and Information Systems

Location-Based Services (LBSs) are one of the most frequently used mobile applications in the modern society. Geo-Indistinguishability (Geo-Ind) is a promising privacy protection model for LBSs since it can provide formal security guarantees for location privacy. However, Geo-Ind undermines the statistical location distribution of users on the LBS server because of perturbed locations, thereby disabling the server to provide distribution-based services (e.g., traffic congestion maps). To overcome this issue, we give a privacy definition, called DistPreserv, to enable the LBS server to acquire valid location distributions while providing users with strict location protection. Then we propose a privacy-preserving LBS scheme …

Towards A Smaller Student: Capacity Dynamic Distillation For Efficient Image Retrieval, Yi Xie, Huaidong Zhang, Xuemiao Xu, Jianqing Zhu, Shengfeng He

Research Collection School Of Computing and Information Systems

Previous Knowledge Distillation based efficient image retrieval methods employ a lightweight network as the student model for fast inference. However, the lightweight student model lacks adequate representation capacity for effective knowledge imitation during the most critical early training period, causing final performance degeneration. To tackle this issue, we propose a Capacity Dynamic Distillation framework, which constructs a student model with editable representation capacity. Specifically, the employed student model is initially a heavy model to fruitfully learn distilled knowledge in the early training epochs, and the student model is gradually compressed during the training. To dynamically adjust the model capacity, our …

Avoiding Starvation Of Arms In Restless Multi-Armed Bandit, Dexun Li, Pradeep Varakantham

Research Collection School Of Computing and Information Systems

Restless multi-armed bandits (RMAB) is a popular framework for optimizing performance with limited resources under uncertainty. It is an extremely useful model for monitoring beneficiaries (arms) and executing timely interventions using health workers (limited resources) to ensure optimal benefit in public health settings. For instance, RMAB has been used to track patients' health and monitor their adherence in tuberculosis settings, ensure pregnant mothers listen to automated calls about good pregnancy practices, etc. Due to the limited resources, typically certain individuals, communities, or regions are starved of interventions, which can potentially have a significant negative impact on the individual/community in the …

Towards Explaining Sequences Of Actions In Multi-Agent Deep Reinforcement Learning Models, Phyo Wai Khaing, Minghong Geng, Budhitama Subagdja, Shubham Pateria, Ah-Hwee Tan

Research Collection School Of Computing and Information Systems

Although Multi-agent Deep Reinforcement Learning (MADRL) has shown promising results in solving complex real-world problems, the applicability and reliability of MADRL models are often limited by a lack of understanding of their inner workings for explaining the decisions made. To address this issue, this paper proposes a novel method for explaining MADRL by generalizing the sequences of action events performed by agents into high-level abstract strategies using a spatio-temporal neural network model. Specifically, an interval-based memory retrieval procedure is developed to generalize the encoded sequences of action events over time into short sequential patterns. In addition, two abstraction algorithms are …

Privacy Please: Power Distance And People’S Responses To Data Breaches Across Countries, Shilpa Madan, Krishna Savani, Constantine S. Katsikeas

Research Collection Lee Kong Chian School Of Business

Information security and data breaches are perhaps the biggest challenges that global businesses face in the digital economy. Although data breaches can cause significant harm to users, businesses, and society, there is significant individual and national variation in people’s responses to data breaches across markets. This research investigates power distance as an antecedent of people’s divergent reactions to data breaches. Eight studies using archival, correlational, and experimental methods find that high power distance makes users more willing to continue patronizing a business after a data breach (Studies 1–3). This is because they are more likely to believe that the business, …

Where Is My Spot? Few-Shot Image Generation Via Latent Subspace Optimization, Chenxi Zheng, Bangzhen Liu, Huaidong Zhang, Xuemiao Xu, Shengfeng He

Research Collection School Of Computing and Information Systems

Image generation relies on massive training data that can hardly produce diverse images of an unseen category according to a few examples. In this paper, we address this dilemma by projecting sparse few-shot samples into a continuous latent space that can potentially generate infinite unseen samples. The rationale behind is that we aim to locate a centroid latent position in a conditional StyleGAN, where the corresponding output image on that centroid can maximize the similarity with the given samples. Although the given samples are unseen for the conditional StyleGAN, we assume the neighboring latent subspace around the centroid belongs to …

Lessons Learned From The Hospital To Home Community Care Program In Singapore And The Supporting Ai Multiple Readmissions Prediction Model, John Abisheganaden, Kheng Hock Lee, Lian Leng Low, Eugene Shum, Han Leong Goh, Christine Gia Lee Ang, Adny An Ta Wee, Steven M. Miller

Research Collection School Of Computing and Information Systems

In a prior practice and policy article published in Healthcare Science, we introduced the deployed application of an artificial intelligence (AI) model to predict longer-term inpatient readmissions to guide community care interventions for patients with complex conditions in the context of Singapore's Hospital to Home (H2H) program that has been operating since 2017. In this follow on practice and policy article, we further elaborate on Singapore's H2H program and care model, and its supporting AI model for multiple readmission prediction, in the following ways: (1) by providing updates on the AI and supporting information systems, (2) by reporting on customer …

Tracing The Twenty-Year Evolution Of Developing Ai For Eye Screening In Singapore: A Master Chronology Of Sidrp, Selena+ And Eyris, Steven M. Miller

Research Collection School Of Computing and Information Systems

This working paper is entirely comprised of a timeline table that begins in 2002 and runs through mid-2023. Across these two decades, this timeline traces the evolutionary development of the following:

• The early Singapore R&D efforts to apply software-based image analysis algorithms and methods to analyse eye retina images for diabetic retinopathy and other eye diseases. This was based on a collaboration between the Singapore Eye Research Institute (SERI) and its parent organization, the Singapore National Eye Centre (SNEC), with faculty from the School of Computing at National University of Singapore.
• The establishment and operation of the Singapore Integrated Diabetic …

Catching The Fast Payments Trend: Optimal Designs And Leadership Strategies Of Retail Payment And Settlement Systems, Zhiling Guo, Dan Ma

Research Collection School Of Computing and Information Systems

Recent financial technologies have enabled fast payments and are reshaping retail payment and settlement systems globally. We developed an analytical model to study the optimal design of a new retail payment system in terms of settlement speed and system capability under both bank and fintech firm heterogeneous participation incentives. We found that three types of payment systems emerge as equilibrium outcomes: batch retail (BR), expedited retail (ER), and real-time retail (RR) payment systems. Although the base value of the payment service positively affects both settlement speed and system capability, the expected liquidity cost negatively impacts settlement speed, and total transaction …

How Does Credit Risk Affect Cost Management Strategies? Evidence On The Initiation Of Credit Default Swap And Sticky Cost Behavior, Jing Dai, Nan Hu, Rong Huang, Yan Yan

Research Collection School Of Computing and Information Systems

In this paper, we examine the effect of credit defaults swaps (CDS) initiation on reference firms' cost management strategies. CDS contracts provide insurance protection for creditors, inducing a shift in bargaining power from borrowers to creditors and an excessive incidence of bankruptcy. Anticipating more intransigent creditors in debt renegotiations and higher bankruptcy risk, CDS firms are incentivized to mitigate risk through decreasing cost stickiness after CDS initiation, as cost stickiness lowers liquidity and triggers early covenant violations. We find that, on average, CDS initiation is associated with a decline in reference firms' cost stickiness. This association is more pronounced for …

Glocal Energy-Based Learning For Few-Shot Open-Set Recognition, Haoyu Wang, Guansong Pang, Peng Wang, Lei Zhang, Wei Wei, Yanning Zhang

Research Collection School Of Computing and Information Systems

Few-shot open-set recognition (FSOR) is a challenging task of great practical value. It aims to categorize a sample to one of the pre-defined, closed-set classes illustrated by few examples while being able to reject the sample from unknown classes. In this work, we approach the FSOR task by proposing a novel energy-based hybrid model. The model is composed of two branches, where a classification branch learns a metric to classify a sample to one of closedset classes and the energy branch explicitly estimates the open-set probability. To achieve holistic detection of openset samples, our model leverages both class-wise and pixelwise …

Class-Incremental Exemplar Compression For Class-Incremental Learning, Zilin Luo, Yaoyao Liu, Bernt Schiele, Qianru Sun

Research Collection School Of Computing and Information Systems

Exemplar-based class-incremental learning (CIL) finetunes the model with all samples of new classes but few-shot exemplars of old classes in each incremental phase, where the "few-shot" abides by the limited memory budget. In this paper, we break this "few-shot" limit based on a simple yet surprisingly effective idea: compressing exemplars by downsampling non-discriminative pixels and saving "many-shot" compressed exemplars in the memory. Without needing any manual annotation, we achieve this compression by generating 0-1 masks on discriminative pixels from class activation maps (CAM). We propose an adaptive mask generation model called class-incremental masking (CIM) to explicitly resolve two difficulties of …

Extracting Class Activation Maps From Non-Discriminative Features As Well, Zhaozheng Chen, Qianru Sun

Research Collection School Of Computing and Information Systems

Extracting class activation maps (CAM) from a classification model often results in poor coverage on foreground objects, i.e., only the discriminative region (e.g., the “head” of “sheep”) is recognized and the rest (e.g., the “leg” of “sheep”) mistakenly as background. The crux behind is that the weight of the classifier (used to compute CAM) captures only the discriminative features of objects. We tackle this by introducing a new computation method for CAM that explicitly captures non-discriminative features as well, thereby expanding CAM to cover whole objects. Specifically, we omit the last pooling layer of the classification model, and perform clustering …

Freestyle Layout-To-Image Synthesis, Han Xue, Zhiwu Huang, Qianru Sun, Li Song, Wenjun Zhang

Research Collection School Of Computing and Information Systems

Typical layout-to-image synthesis (LIS) models generate images for a close set of semantic classes, e.g., 182 common objects in COCO-Stuff. In this work, we explore the freestyle capability of the model, i.e., how far can it generate unseen semantics (e.g., classes, attributes, and styles) onto a given layout, and call the task Freestyle LIS (FLIS). Thanks to the development of large-scale pre-trained language-image models, a number of discriminative models (e.g., image classification and object detection) trained on limited base classes are empowered with the ability of unseen class prediction. Inspired by this, we opt to leverage large-scale pre-trained text-to-image diffusion …

3d Dental Biometrics: Transformer-Based Dental Arch Extraction And Matching, Zhiyuan Zhang, Zhong Xin

Research Collection School Of Computing and Information Systems

The dental arch is a significant anatomical feature that is crucial in assessing tooth arrangement and configuration and has a potential for human identification in biometrics and digital forensic dentistry. In a previous study, we proposed an auto pose-invariant arch feature extraction Radial Ray Algorithm (RRA) and a matching framework [1] based solely on 3D dental geometry. To enhance the identification accuracy and speed of our previous work, we propose in this study a transformer architecture that can extract dental keypoints by encoding both local and global features. The dental arch is then constructed through robust interpolation of the dental …

Improving Quantal Cognitive Hierarchy Model Through Iterative Population Learning, Yuhong Xu, Shih-Fen Cheng, Xinyu Chen

Research Collection School Of Computing and Information Systems

In this paper, we propose to enhance the state-of-the-art quantal cognitive hierarchy (QCH) model with iterative population learning (IPL) to estimate the empirical distribution of agents’ reasoning levels and fit human agents’ behavioral data. We apply our approach to a real-world dataset from the Swedish lowest unique positive integer (LUPI) game and show that our proposed approach outperforms the theoretical Poisson Nash equilibrium predictions and the QCH approach by 49.8% and 46.6% in Wasserstein distance respectively. Our approach also allows us to explicitly measure an agent’s reasoning level distribution, which is not previously possible.

Strategic Planning For Flexible Agent Availability In Large Taxi Fleets, Rajiv Ranjan Kumar, Pradeep Varakantham, Shih-Fen Cheng

Research Collection School Of Computing and Information Systems

In large scale multi-agent systems like taxi fleets, individual agents (taxi drivers) are self interested (maximizing their own profits) and this can introduce inefficiencies in the system. One such inefficiency is with regards to the "required" availability of taxis at different time periods during the day. Since a taxi driver can work for limited number of hours in a day (e.g., 8-10 hours in a city like Singapore), there is a need to optimize the specific hours, so as to maximize individual as well as social welfare. Technically, this corresponds to solving a large scale multi-stage selfish routing game with …

How To Resuscitate A Sick Vm In The Cloud, Xuhua Ding

Research Collection School Of Computing and Information Systems

A guest virtual machine in a cloud platform may fall “sick” when its kernel encounters a fatal low-level bug or is subverted by an adversary. The VM owner is hence likely to lose her control over it due to a kernel hang or being denied of remote accesses. While the VM can be rebooted with the assistance from the cloud server, the owner not only faces service disruption but also is left with no opportunity to make an in-depth diagnosis and forensics on the spot, not to mention a live rectification. Currently, the cloud service provider has neither incentive nor …

Semantic Scene Completion With Cleaner Self, Fengyun Wang, Dong Zhang, Hanwang Zhang, Jinhui Tang, Qianru Sun

Research Collection School Of Computing and Information Systems

Semantic Scene Completion (SSC) transforms an image of single-view depth and/or RGB 2D pixels into 3D voxels, each of whose semantic labels are predicted. SSC is a well-known ill-posed problem as the prediction model has to “imagine” what is behind the visible surface, which is usually represented by Truncated Signed Distance Function (TSDF). Due to the sensory imperfection of the depth camera, most existing methods based on the noisy TSDF estimated from depth values suffer from 1) incomplete volumetric predictions and 2) confused semantic labels. To this end, we use the ground-truth 3D voxels to generate a perfect visible surface, …

Unbiased Multiple Instance Learning For Weakly Supervised Video Anomaly Detection, Hui Lyu, Zhongqi Yue, Qianru Sun, Bin Luo, Zhen Cui, Hanwang Zhang

Research Collection School Of Computing and Information Systems

Weakly Supervised Video Anomaly Detection (WSVAD) is challenging because the binary anomaly label is only given on the video level, but the output requires snippet-level predictions. So, Multiple Instance Learning (MIL) is prevailing in WSVAD. However, MIL is notoriously known to suffer from many false alarms because the snippet-level detector is easily biased towards the abnormal snippets with simple context, confused by the normality with the same bias, and missing the anomaly with a different pattern. To this end, we propose a new MIL framework: Unbiased MIL (UMIL), to learn unbiased anomaly features that improve WSVAD. At each MIL training …

Livoauth: Liveness Detection In Voiceprint Authentication With Random Challenges And Detection Modes, Rui Zhang, Zheng Yan, Xueru Wang, Robert H. Deng

Research Collection School Of Computing and Information Systems

Voiceprint authentication provides great convenience to users in many application scenarios. However, it easily suffers from spoofing attacks including speech synthesis, speech conversion, and speech replay. Liveness detection is an effective way to resist these attacks. But existing methods suffer from many disadvantages, such as extra deployment costs due to precise data collection, environmental disturbance, high computational overhead, and operational complexity. A uniform platform that can offer voiceprint authentication as a service (VAaS) over the cloud is also lacked. Hence, it is imperative to design an economic and effective method for liveness detection in voiceprint authentication. In this article, we …

Evading Deepfake Detectors Via Adversarial Statistical Consistency, Yang Hou, Qing Guo, Yihao Huang, Xiaofei Xie, Lei Ma, Jianjun Zhao

Research Collection School Of Computing and Information Systems

In recent years, as various realistic face forgery techniques known as DeepFake improves by leaps and bounds, more and more DeepFake detection techniques have been proposed. These methods typically rely on detecting statistical differences between natural (i.e., real) and DeepFake-generated images in both spatial and frequency domains. In this work, we propose to explicitly minimize the statistical differences to evade state-of-the-art DeepFake detectors. To this end, we propose a statistical consistency attack (StatAttack) against DeepFake detectors, which contains two main parts. First, we select several statistical-sensitive natural degradations (i.e., exposure, blur, and noise) and add them to the fake images …

The Bemi Stardust: A Structured Ensemble Of Binarized Neural Networks, Ambrogio Maria Bernardelli, Stefano Gualandi, Hoong Chuin Lau, Simone Milanesi

Research Collection School Of Computing and Information Systems

Binarized Neural Networks (BNNs) are receiving increasing attention due to their lightweight architecture and ability to run on low-power devices, given the fact that they can be implemented using Boolean operations. The state-of-the-art for training classification BNNs restricted to few-shot learning is based on a Mixed Integer Programming (MIP) approach. This paper proposes the BeMi ensemble, a structured architecture of classification-designed BNNs based on training a single BNN for each possible pair of classes and applying a majority voting scheme to predict the final output. The training of a single BNN discriminating between two classes is achieved by a MIP …