Physical Sciences and Mathematics Commons^™

Online Deep Learning: Learning Deep Neural Networks On The Fly, Doyen Sahoo, Hong Quang Pham, Jing Lu, Steven C. H. Hoi

Research Collection School Of Computing and Information Systems

Deep Neural Networks (DNNs) are typically trained by backpropagation in a batch setting, requiring the entire training data to be made available prior to the learning task. This is not scalable for many real-world scenarios where new data arrives sequentially in a stream. We aim to address an open challenge of “Online Deep Learning” (ODL) for learning DNNs on the fly in an online setting. Unlike traditional online learning that often optimizes some convex objective function with respect to a shallow model (e.g., a linear/kernel-based hypothesis), ODL is more challenging as the optimization objective is non-convex, and regular DNN with …

Analysis Of Public Transportation Patterns In A Densely Populated City With Station-Based Shared Bikes, Di Wang, Evan Wu, Ah-Hwee Tan

Research Collection School Of Computing and Information Systems

Densely populated cities face great challenges of high transportation demand and limited physical space. Thus, in these cities, the public transportation system is heavily relied on. Conventional public transportation modes such as bus, taxi and subway have been globally deployed over the past century. In the last decade, a new type of public transportation mode, shared bike, emerged in many cities. These shared bikes are deployed by either government-regulated or profit-driven companies and are either station-based or station-less. Nonetheless, all of them are designed to better solve the last-mile problem in densely populated cities as complements to the conventional public …

Hybrid Recommender For Online Petitions With Social Network And Psycholinguistic Features, Ahmed Elnoshokaty

Masters Theses & Doctoral Dissertations

The online petition has become one of the most important channels of civic participation. Most of the state-of-the-art online platforms, however, tend to use simple indicators (such as popularity) to rank petitions, hence creating a situation where the most popular petitions dominate the rank and attract most people’s attention. For the petitions which focus on specific issues, they are often in a disadvantageous position on the list. For example, a petition for local environment problem may not be seen by many people who are really concerned with it, simply because it takes multiple pages to reach it. Therefore, the simple …

A Driver Guidance System For Taxis In Singapore, Shashi Shekhar Jha, Shih-Fen Cheng, Meghna Lowalekar, Nicholas Wong, Rishikeshan Rajendram, Pradeep Varakantham, Nghia Troung Troung, Firmansyah Bin Abd Rahman

Research Collection School Of Computing and Information Systems

Traditional taxi fleet operators world-over have been facing intense competitions from various ride-hailing services such as Uber and Grab.Based on our studies on the taxi industry in Singapore, we see that the emergence of Uber and Grab in the ride-hailing market has greatly impacted the taxi industry: the average daily taxi ridership for the past two years has been falling continuously, by close to 20% in total. In this work, we discuss how efficient real-time data analytics and large-scale multiagent optimization technology could help taxi drivers compete against more technologically advanced service platforms. Our system has been in field trial …

Face Detection Using Deep Learning: An Improved Faster Rcnn Approach, Xudong Sun, Pengcheng Wu, Steven C. H. Hoi

Research Collection School Of Computing and Information Systems

In this paper, we present a new face detection scheme using deep learning and achieve the state-of-the-art detection performance on the well-known FDDB face detection benchmark evaluation. In particular, we improve the state-of-the-art Faster RCNN framework by combining a number of strategies, including feature concatenation, hard negative mining, multi-scale training, model pre-training, and proper calibration of key parameters. As a consequence, the proposed scheme obtained the state-of-the-art face detection performance and was ranked as one of the best models in terms of ROC curves of the published methods on the FDDB benchmark

Distributed K-Nearest Neighbor Queries In Metric Spaces, Xin Ding, Yuanliang Zhang, Lu Chen, Yunjun Gao, Baihua Zheng

Research Collection School Of Computing and Information Systems

Metric k nearest neighbor (MkNN) queries have applications in many areas such as multimedia retrieval, computational biology, and location-based services. With the growing volumes of data, a distributed method is required. In this paper, we propose an Asynchronous Metric Distributed System (AMDS), which uniformly partitions the data with the pivot-mapping technique to ensure the load balancing, and employs publish/subscribe communication model to asynchronously process large scale of queries. The employment of asynchronous processing model also improves robustness and efficiency of AMDS. In addition, we develop an efficient estimation based MkNN method using AMDS to improve the query efficiency. Extensive experiments …

Rumor Detection On Twitter With Tree-Structured Recursive Neural Networks, Jing Ma, Wei Gao, Kam-Fai Wong

Research Collection School Of Computing and Information Systems

Sentiment expression in microblog posts can be affected by user’s personal character, opinion bias, political stance and so on. Most of existing personalized microblog sentiment classification methods suffer from the insufficiency of discriminative tweets for personalization learning. We observed that microblog users have consistent individuality and opinion bias in different languages. Based on this observation, in this paper we propose a novel user-attention-based Convolutional Neural Network (CNN) model with adversarial cross-lingual learning framework. The user attention mechanism is leveraged in CNN model to capture user’s language-specific individuality from the posts. Then the attention-based CNN model is incorporated into a novel …

Autonomous Agents In Snake Game Via Deep Reinforcement Learning, Zhepei Wei, Di Wang, Ming Zhang, Ah-Hwee Tan, Chunyan Miao, You Zhou

Research Collection School Of Computing and Information Systems

Since DeepMind pioneered a deep reinforcement learning (DRL) model to play the Atari games, DRL has become a commonly adopted method to enable the agents to learn complex control policies in various video games. However, similar approaches may still need to be improved when applied to more challenging scenarios, where reward signals are sparse and delayed. In this paper, we develop a refined DRL model to enable our autonomous agent to play the classical Snake Game, whose constraint gets stricter as the game progresses. Specifically, we employ a convolutional neural network (CNN) trained with a variant of Q-learning. Moreover, we …

Is Information Systems Misuse Always Bad? A New Perspective On Is Misuse In Hospitals Under The Context Of Disasters, Dheyaaldin Alsalman

Masters Theses & Doctoral Dissertations

Although the extant literature has investigated how individuals engage in inappropriate behaviors based on the rational choice theory (RCT) (e.g., computer misconduct), the neutralization theory (e.g., IS security policies violation), and workarounds under normal situations, it has given little consideration to how individuals are involved in misuse of information systems with a good intention under the context of disasters. To fill this research gap, we propose a selfless misuse model, which offers a theoretical explanation for the concept of individuals’ selfless misuse intention under uncertainty caused by disasters. In this study, we show why employees make decisions to misuse the …

Online Active Learning With Expert Advice, Shuji Hao, Peiying Hu, Peilin Zhao, Steven C. H. Hoi, Chunyan Miao

Research Collection School Of Computing and Information Systems

In literature, learning with expert advice methods usually assume that a learner always obtain the true label of every incoming training instance at the end of each trial. However, in many real-world applications, acquiring the true labels of all instances can be both costly and time consuming, especially for large-scale problems. For example, in the social media, data stream usually comes in a high speed and volume, and it is nearly impossible and highly costly to label all of the instances. In this article, we address this problem with active learning with expert advice, where the ground truth of an …

Efficient Representative Subset Selection Over Sliding Windows, Yanhao Wang, Yuchen Li, Kian-Lee Tan

Research Collection School Of Computing and Information Systems

Representative subset selection (RSS) is an important tool for users to draw insights from massive datasets. Existing literature models RSS as submodular maximization to capture the "diminishing returns" property of representativeness, but often only has a single constraint, which limits its applications to many real-world problems. To capture the recency issue and support various constraints, we formulate dynamic RSS as maximizing submodular functions subject to general d -knapsack constraints (SMDK) over sliding windows. We propose a KnapWindow framework (KW) for SMDK. KW utilizes KnapStream (KS) for SMDK in append-only streams as a subroutine. It maintains a sequence of checkpoints and …

Deeptravel: A Neural Network Based Travel Time Estimation Model With Auxiliary Supervision, Hanyuan Zhang, Hao Wu, Weiwei Sun, Baihua Zheng

Research Collection School Of Computing and Information Systems

Estimating the travel time of a path is of great importance to smart urban mobility. Existing approaches are either based on estimating the time cost of each road segment or designed heuristically in a non-learning-based way. The former is not able to capture many cross-segment complex factors while the latter fails to utilize the existing abundant temporal labels of the data, i.e., the time stamp of each trajectory point. In this paper, we leverage on new development of deep neural networks and propose a novel auxiliary supervision model, namely DeepTravel, that can automatically and effectively extract different features, as well …

Modeling Contemporaneous Basket Sequences With Twin Networks For Next-Item Recommendation, Duc Trong Le, Hady W. Lauw, Yuan Fang

Research Collection School Of Computing and Information Systems

Our interactions with an application frequently leave a heterogeneous and contemporaneous trail of actions and adoptions (e.g., clicks, bookmarks, purchases). Given a sequence of a particular type (e.g., purchases)-- referred to as the target sequence, we seek to predict the next item expected to appear beyond this sequence. This task is known as next-item recommendation. We hypothesize two means for improvement. First, within each time step, a user may interact with multiple items (a basket), with potential latent associations among them. Second, predicting the next item in the target sequence may be helped by also learning from another supporting sequence …

Taxis Strike Back: A Field Trial Of The Driver Guidance System, Shih-Fen Cheng, Shashi Shekhar Jha, Rishikeshan Rajendram

Research Collection School Of Computing and Information Systems

Traditional taxi fleet operators world-over have been facing intense competitions from various ride-hailing services such as Uber and Grab (specific to the Southeast Asia region). Based on our studies on the taxi industry in Singapore, we see that the emergence of Uber and Grab in the ride-hailing market has greatly impacted the taxi industry: the average daily taxi ridership for the past two years has been falling continuously, by close to 20% in total. In this work, we discuss how efficient real-time data analytics and large-scale multi-agent optimization technology could potentially help taxi drivers compete against more technologically advanced service …

Striving To Earn More: A Survey Of Work Strategies And Tool Use Among Crowd Workers, Toni Kaplan, Susumu Saito, Kotaro Hara, Jeffrey P. Bigham

Research Collection School Of Computing and Information Systems

Earning money is a primary motivation for workers on Amazon Mechanical Turk, but earning a good wage is difficult because work that pays well is not easily identified and can be time-consuming to find. We explored the strategies that both low- and high-earning workers use to find and complete tasks via a survey of 360 workers. Nearly all workers surveyed had earning money as their primary goal, and workers used many of the same tools (browser extensions and scripts) and strategies in an attempt to earn more money, regardless of earning level. However, high-earning workers used more tools, were more …

Knowledge-Aware Attentive Neural Network For Ranking Question Answer Pairs, Ying Shen, Yang Deng, Min Yang, Yaliang Li, Nan Du, Wei Fan, Kai Lei

Research Collection School Of Computing and Information Systems

Ranking question answer pairs has attracted increasing attention recently due to its broad applications such as information retrieval and question answering (QA). Significant progresses have been made by deep neural networks. However, background information and hidden relations beyond the context, which play crucial roles in human text comprehension, have received little attention in recent deep neural networks that achieve the state of the art in ranking QA pairs. In the paper, we propose KABLSTM, a Knowledge-aware Attentive Bidirectional Long Short-Term Memory, which leverages external knowledge from knowledge graphs (KG) to enrich the representational learning of QA sentences. Specifically, we develop …

An Effectual Approach For The Development Of Novel Applications On Digital Platforms, Onkar Shamrao Malgonde

USF Tampa Graduate Theses and Dissertations

The development of novel software applications on digital platforms differs from traditional software development and provides unique challenges to the software development manager and team. Application producers must achieve application-platform match, application-market match, value propositions exceeding platform’s core value propositions, and novelty. These desired properties support a new vision of the software development team as entrepreneurs with a goal of developing novel applications on digital platforms. Digital platforms are characterized by an uncertain, risky, and resource-constrained environment, where existing approaches—plan-driven, ad-hoc, and controlled-flexible—have limited applicability. Building on the theoretical basis of the theory of effectuation from the entrepreneurship domain, this …

Querying Large Databases, Nathan Beneke

Scholarly Horizons: University of Minnesota, Morris Undergraduate Journal

This paper investigates two approaches to improving query times on large relational databases. The first technique capitalizes on the knowledge of a database's structures and properties one typically has. This technique can execute some queries exactly in a constant, bounded amount of time. When this technique cannot be used to exactly execute a query we show how it can still be used to drastically lower the run-time on the query while getting a good approximation of the exact result. We also discuss the complexity of deciding whether a query is evaluable in this way, both theoretically and practically. The second …

The Effect Of Endgame Tablebases On Modern Chess Engines, Christopher D. Peterson

Computer Engineering

Modern chess engines have the ability to augment their evaluation by using massive tables containing billions of positions and their memorized solutions. This report examines the importance of these tables to better understand the circumstances under which they should be used. The analysis conducted in this paper empirically examines differences in size and speed of memorized positions and their impacts on engine strength. Using this technique, situations where memorized tables improve play (and situations where they do not) are discovered.

Augmented Personalized Health: Using Semantically Integrated Multimodal Data For Patient Empowered Health Management Strategies, Amit P. Sheth, Hong Y. Yip, Utkarshani Jaimini, Dipesh Kadariya, Vaikunth Sridharan, R. Venkataramanan, Tanvi Banerjee, Krishnaprasad Thirunarayan, Maninder Kalra

Kno.e.sis Publications

Healthcare as we know it is in the process of going through a massive change from:

1. Episodic to continuous

2. Disease-focused to wellness and quality of life focused

3. Clinic-centric to anywhere a patient is

4. Clinician controlled to patient empowered

5. Being driven by limited data to 360-degree, multimodal personal-public-population physical-cyber-social big data-driven URL: https://mhealth.md2k.org/2018-tech-showcase-home

Deaddrop: Message Passing Without Metadata Leakage, Davis Mike Arndt

Computer Science and Software Engineering

Even when network data is encrypted, observers can make inferences about content based on collected metadata. DeadDrop is an exploratory API designed to protect the metadata of a conversation from both outside observers and the facilitating server. To do so, DeadDrop servers are passed no recipient address, instead relying upon the recipient to check for messages of their own volition. In addition, the recipient downloads a copy of every encrypted message on the server to prevent even the server from knowing to whom each message is intended. To these purposes, DeadDrop is mostly successful. However, it does not obscure all …

An Economic Analysis Of Disintermediation On Crowdfunding Platforms, Jianqing Chen, Ling Ge, Zhiling Guo

Research Collection School Of Computing and Information Systems

Prosocial crowdfunding platforms can work through direct peer-to-peer (P2P) lending or through intermediaries, incurring different costs to borrowers and lenders. This study investigates the incentives of lenders and borrowers’ and how they would choose between the two types of platforms. We model the intermediary as a profit maximizer who filters projects, provides high quality borrowers with access to the platform, and ensures repayment rate to lenders. Our initial findings suggest that the introduction of direct P2P lending platform enables the intermediary to reduce its interest rate and to raise its screening threshold on the intermediated platform. The P2P lending platform …

A Proposal For A Decentralized Liquidity Savings Mechanism With Side Payments, Adam Fugal, Rodney Garratt, Zhiling Guo, Dave Hudson

Research Collection School Of Computing and Information Systems

In most countries, the central bank provides the medium to physically settle the smallest payments (cash) and the means to electronically settle the largest payments, which typically are wholesale payments between banks. For the latter purpose the central bank usually operates a system through which banks can settle payments in central bank money. Historically, interbank payments were settled via (end of day) netting systems, but as volumes and values increased central banks became worried about the risks inherent in deferred net settlement systems, so most central banks opted for the implementation of a Real Time Gross Settlement (RTGS) system. With …

Covariance Pooling For Facial Expression Recognition, D. Acharya, Zhiwu Huang, D. Paudel, Gool L. Van

Research Collection School Of Computing and Information Systems

Classifying facial expressions into different categories requires capturing regional distortions of facial landmarks. We believe that second-order statistics such as covariance is better able to capture such distortions in regional facial features. In this work, we explore the benefits of using a manifold network structure for covariance pooling to improve facial expression recognition. In particular, we first employ such kind of manifold networks in conjunction with traditional convolutional networks for spatial pooling within individual image feature maps in an end-to-end deep learning manner. By doing so, we are able to achieve a recognition accuracy of 58.14% on the validation set …

Region-Aware Reflection Removal With Unified Content And Gradient Priors, Renjie Wan, Boxin Shi, Ling-Yu Duan, Ah-Hwee Tan, Wen Gao, Alex C. Kot

Research Collection School Of Computing and Information Systems

Removing the undesired reflections in images taken through the glass is of broad application to various image processing and computer vision tasks. Existing single image-based solutions heavily rely on scene priors such as separable sparse gradients caused by different levels of blur, and they are fragile when such priors are not observed. In this paper, we notice that strong reflections usually dominant a limited region in the whole image, and propose a region-aware reflection removal approach by automatically detecting and heterogeneously processing regions with and without reflections. We integrate content and gradient priors to jointly achieve missing contents restoration, as …

From 2,772 Segments To Five Personas: Summarizing A Diverse Online Audience By Generating Culturally Adapted Personas, Joni Salminen, Sercan Sengun, Haewoon Kwak, Bernard J. Jansen, Jisun An, Soon-Gyu Jung, Sarah Vieweg, D. Fox Harrell

Research Collection School Of Computing and Information Systems

Understanding users in the era of social media is challenging, requiring organizations to adopt novel computation-aided approaches. To exemplify such an approach, we retrieved information on millions of interactions with YouTube video content from a major Middle Eastern media outlet, to automatically generate personas that capture how different audience segments interact with thousands of individual content pieces. Then, we used qualitative data to provide additional insights into the automatically generated persona profiles. Our findings provide insights into social media usage in the Middle East and demonstrate the application of a novel methodology that generates culturally adapted personas of social media …

Assessing The Accuracy Of Four Popular Face Recognition Tools For Inferring Gender, Age, And Race, Soon-Gyu Jung, Jisun An, Haewoon Kwak, Joni Salminen, Bernard J. Jansen

Research Collection School Of Computing and Information Systems

In this research, we evaluate four widely used face detection tools, which are Face++, IBM Bluemix Visual Recognition, AWS Rekognition, and Microsoft Azure Face API, using multiple datasets to determine their accuracy in inferring user attributes, including gender, race, and age. Results show that the tools are generally proficient at determining gender, with accuracy rates greater than 90%, except for IBM Bluemix. Concerning race, only one of the four tools provides this capability, Face++, with an accuracy rate of greater than 90%, although the evaluation was performed on a high-quality dataset. Inferring age appears to be a challenging problem, as …

Automatically Conceptualizing Social Media Analytics Data Via Personas, Jung S.G., Salminen J., An J., Kwak H., Jansen B.J.

Research Collection School Of Computing and Information Systems

Social media analytics is insightful but can also be difficult to use within organizations. To address this, we present Automatic Persona Generation (APG), a system and methodology for quantitatively generating personas using large amounts of online social media data. The APG system is operational, deployed in a pilot version with several organizations in multiple industry verticals. APG uses a robust web and stable back-end database framework to process tens of millions of user interactions with thousands of online digital products on multiple social media platforms, including Facebook and YouTube. APG identifies both distinct and impactful audience segments for an organization …

How Does Developer Interaction Relate To Software Quality? An Examination Of Product Development Data, Subhajit Datta

Research Collection School Of Computing and Information Systems

Industrial software systems are being increasingly developed by large and distributed teams. Tools like collaborative development environments (CDE) are used to facilitate interaction between members of such teams, with the expectation that social factors around the interaction would facilitate team functioning. In this paper, we first identify typically social characteristics of interaction in a software development team: reachability, connection, association, and clustering. We then examine how these factors relate to the quality of software produced by a team, in terms of the number of defects, through an empirical study of 70+ teams, involving 900+ developers in total, spread across 30+ …

Applying Spatial Database Techniques To Other Domains: A Case Study On Top-K And Computational Geometric Operators, Kyriakos Mouratidis

Research Collection School Of Computing and Information Systems

In this seminar, we will explore how processing rich spatial data is not the only practical (and research-wise promising) application domain for traditional spatial database techniques. An equally promising direction, possibly with low-hanging fruits for research innovation, may be to apply the spatial data management expertise of our community to non-spatial types of queries, and to extend standard, more theoretical operators to large scale datasets with the objective of practical solutions (as opposed to favorable asymptotic complexity alone). As a case study, we will review spatial database work on top-k-related operators (i.e., non-spatial problems) and how it integrates fundamental computational …