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

Physical Sciences and Mathematics Commons™

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

Research Collection School Of Computing and Information Systems

Discipline
Keyword
Publication Year
File Type

Articles 5281 - 5310 of 6891

Full-Text Articles in Physical Sciences and Mathematics

Packed, Printable, And Polymorphic Return-Oriented Programming, Kangjie Lu, Dabi Zou, Weiping Wen, Debin Gao Sep 2011

Packed, Printable, And Polymorphic Return-Oriented Programming, Kangjie Lu, Dabi Zou, Weiping Wen, Debin Gao

Research Collection School Of Computing and Information Systems

Return-oriented programming (ROP) is an attack that has been shown to be able to circumvent W ⊕ X protection. However, it was not clear if ROP can be made as powerful as non-ROP malicious code in other aspects, e.g., be packed to make static analysis difficult, be printable to evade non-ASCII filtering, be polymorphic to evade signature-based detection, etc. Research in these potential advances in ROP is important in designing counter-measures. In this paper, we show that ROP code could be packed, printable, and polymorphic. We demonstrate this by proposing a packer that produces printable and polymorphic ROP code. It …

Go to article

Linear Obfuscation To Combat Symbolic Execution, Zhi Wang, Jiang Ming, Chunfu Jia, Debin Gao Sep 2011

Linear Obfuscation To Combat Symbolic Execution, Zhi Wang, Jiang Ming, Chunfu Jia, Debin Gao

Research Collection School Of Computing and Information Systems

Trigger-based code (malicious in many cases, but not necessarily) only executes when specific inputs are received. Symbolic execution has been one of the most powerful techniques in discovering such malicious code and analyzing the trigger condition. We propose a novel automatic malware obfuscation technique to make analysis based on symbolic execution difficult. Unlike previously proposed techniques, the obfuscated code from our tool does not use any cryptographic operations and makes use of only linear operations which symbolic execution is believed to be good in analyzing. The obfuscated code incorporates unsolved conjectures and adds a simple loop to the original code, …

Go to article

When Recommendation Meets Mobile: Contextual And Personalised Recommendation On The Go, Jinfeng Zhuang, Tao Mei, Steven C. H. Hoi, Ying-Qing Xu, Shipeng Li Sep 2011

When Recommendation Meets Mobile: Contextual And Personalised Recommendation On The Go, Jinfeng Zhuang, Tao Mei, Steven C. H. Hoi, Ying-Qing Xu, Shipeng Li

Research Collection School Of Computing and Information Systems

Mobile devices are becoming ubiquitous. People use their phones as a personal concierge discovering and making decisions anywhere and anytime. Understanding user intent on the go therefore becomes important for task completion on the phone. While existing efforts have predominantly focused on understanding the explicit user intent expressed by a textual or voice query, this paper presents an approach to context-aware and personalized entity recommendation which understands the implicit intent without any explicit user input on the phone. The approach, highly motivated from a large-scale mobile click-through analysis, is able to rank both the entity types and the entities within …

Go to article

On Detection Of Erratic Arguments, Jin Han, Qiang Yan, Robert H. Deng, Debin Gao Sep 2011

On Detection Of Erratic Arguments, Jin Han, Qiang Yan, Robert H. Deng, Debin Gao

Research Collection School Of Computing and Information Systems

Due to the erratic nature, the value of a function argument in one normal program execution could become illegal in another normal execution context. Attacks utilizing such erratic arguments are able to evade detections as fine-grained context information is unavailable in many existing detection schemes. In order to obtain such fine-grained context information, a precise model on the internal program states has to be built, which is impractical especially monitoring a closed source program alone. In this paper, we propose an intrusion detection scheme which builds on two diverse programs providing semantically-close functionality. Our model learns underlying semantic correlation of …

Go to article

Spectral Geometry Image: Image Based 3d Models For Digital Broadcasting Applications, Boon Seng Chew, Lap Pui Chau, Ying He, Dayong Wang, Steven C. H. Hoi Sep 2011

Spectral Geometry Image: Image Based 3d Models For Digital Broadcasting Applications, Boon Seng Chew, Lap Pui Chau, Ying He, Dayong Wang, Steven C. H. Hoi

Research Collection School Of Computing and Information Systems

The use of 3D models for progressive transmission and broadcasting applications is an interesting challenge due to the nature and complexity of such content. In this paper, a new image format for the representation of 3D progressive model is proposed. The powerful spectral analysis is combined with the state of art Geometry Image(GI) to encode static 3D models into spectral geometry images(SGI) for robust 3D shape representation. Based on the 3D model's surface characteristics, SGI separated the geometrical image into low and high frequency layers to achieve effective Level of Details(LOD) modeling. For SGI, the connectivity data of the model …

Go to article

Mining Top-K Large Structural Patterns In A Massive Network, Feida Zhu, Qiang Qu, David Lo, Xifeng Yan, Jiawei Han, Philip S. Yu Sep 2011

Mining Top-K Large Structural Patterns In A Massive Network, Feida Zhu, Qiang Qu, David Lo, Xifeng Yan, Jiawei Han, Philip S. Yu

Research Collection School Of Computing and Information Systems

With ever-growing popularity of social networks, web and bio-networks, mining large frequent patterns from a single huge network has become increasingly important. Yet the existing pattern mining methods cannot offer the efficiency desirable for large pattern discovery. We propose Spider- Mine, a novel algorithm to efficiently mine top-K largest frequent patterns from a single massive network with any user-specified probability of 1 − ϵ. Deviating from the existing edge-by-edge (i.e., incremental) pattern-growth framework, SpiderMine achieves its efficiency by unleashing the power of small patterns of a bounded diameter, which we call “spiders”. With the spider structure, our approach adopts a …

Go to article

An Efficient Adaptive Vortex Particle Method For Real-Time Smoke Simulation, Shengfeng He, Hon-Cheng Wong, Un-Hong Wong Sep 2011

An Efficient Adaptive Vortex Particle Method For Real-Time Smoke Simulation, Shengfeng He, Hon-Cheng Wong, Un-Hong Wong

Research Collection School Of Computing and Information Systems

Smoke simulation is one of the interesting topics in computer animation and it usually involves turbulence generation. Efficient generation of realistic turbulent flows becomes one of the challenges in smoke simulation. Vortex particle method, which is a hybrid method that combines grid-based and particle-based approaches, is often used for generating turbulent details. However, it may cause irrational artifacts due to its initial condition and vorticity forcing approach used. In this paper, a new vorticity forcing approach based on the spatial adaptive vorticity confinement is proposed to address this problem. In this approach, the spatial adaptive vorticity confinement force varies with …

Go to article

Effective Communication Of Software Development Knowledge Through Community Portals, Christoph Treude, Margaret-Anne Storey Sep 2011

Effective Communication Of Software Development Knowledge Through Community Portals, Christoph Treude, Margaret-Anne Storey

Research Collection School Of Computing and Information Systems

Knowledge management plays an important role in many software organizations. Knowledge can be captured and distributed using a variety of media, including traditional help files and manuals, videos, technical articles, wikis, and blogs. In recent years, web-based community portals have emerged as an important mechanism for combining various communication channels. However, there is little advice on how they can be effectively deployed in a software project.In this paper, we present a first study of a community portal used by a closed source software project. Using grounded theory, we develop a model that characterizes documentation artifacts along several dimensions, such as …

Go to article

Tamper Detection In The Epc Network Using Digital Watermarking, Shui-Hua Han, Chao-Hsien Chu, Zongwei Luo Sep 2011

Tamper Detection In The Epc Network Using Digital Watermarking, Shui-Hua Han, Chao-Hsien Chu, Zongwei Luo

Research Collection School Of Computing and Information Systems

One of the most relevant problems in radio frequency identification (RFID) technology is the lack of security measures in its wireless communication channel between the reader and tag. This article analyzes potential data tampering threats in the electronic product code (EPC) network and proposes solutions using fragile watermarking technologies.

Go to article

Structural Analysis Of The Hot Spots In The Binding Between H1n1 Ha And The 2di Antibody: Do Mutations Of H1n1 From 1918 To 2009 Affect Much On This Binding?, Qian Liu, Steven C. H. Hoi, Chinh T. T. Su, Zhenhua Li, Chee-Keong Kwoh, Limsoon Wong, Jinyan Li Sep 2011

Structural Analysis Of The Hot Spots In The Binding Between H1n1 Ha And The 2di Antibody: Do Mutations Of H1n1 From 1918 To 2009 Affect Much On This Binding?, Qian Liu, Steven C. H. Hoi, Chinh T. T. Su, Zhenhua Li, Chee-Keong Kwoh, Limsoon Wong, Jinyan Li

Research Collection School Of Computing and Information Systems

Worldwide and substantial mortality caused by the 2009 H1N1 influenza A has stimulated a new surge of research on H1N1 viruses. An epitope conservation has been learned in the HA1 protein that allows antibodies to cross-neutralize both 1918 and 2009 H1N1. However, few works have thoroughly studied the binding hot spots in those two antigen–antibody interfaces which are responsible for the antibody cross-neutralization. We apply predictive methods to identify binding hot spots at the epitope sites of the HA1 proteins and at the paratope sites of the 2D1 antibody. We find that the six mutations at the HA1's epitope from …

Go to article

Driverguard: A Fine-Grained Protection On I/O Flow, Yueqiang Cheng, Xuhua Ding, Robert H. Deng Sep 2011

Driverguard: A Fine-Grained Protection On I/O Flow, Yueqiang Cheng, Xuhua Ding, Robert H. Deng

Research Collection School Of Computing and Information Systems

Most commodity peripheral devices and their drivers are geared to achieve high performance with security functions being opted out. The absence of security measures invites attacks on the I/O data and consequently threats those applications feeding on them, such as biometric authentication. In this paper, we present the design and implementation of DriverGuard, a hypervisor based protection mechanism which dynamically shields I/O flows such that I/O data are not exposed to the malicious kernel. Our design leverages a composite of cryptographic and virtualization techniques to achieve fine-grained protection. DriverGuard is lightweight as it only needs to protect around 2% of …

Go to article

Structural Complexity And Programmer Team Strategy: An Experimental Test, Narayan Ramasubbu, Chris F. Kemerer, Jeff Min Teck Hong Sep 2011

Structural Complexity And Programmer Team Strategy: An Experimental Test, Narayan Ramasubbu, Chris F. Kemerer, Jeff Min Teck Hong

Research Collection School Of Computing and Information Systems

This study develops and empirically tests the idea that the impact of structural complexity on perfective maintenance of object-oriented software is significantly determined by the team strategy of programmers (independent or collaborative). We analyzed two key dimensions of software structure, coupling and cohesion, with respect to the maintenance effort and the perceived ease-of-maintenance by pairs of programmers. Hypotheses based on the distributed cognition and task interdependence theoretical frameworks were tested using data collected from a controlled lab experiment employing professional programmers. The results show a significant interaction effect between coupling, cohesion, and programmer team strategy on both maintenance effort and …

Go to article

Certificateless Cryptography With Kgc Trust Level 3, Guomin Yang, Chik How Tan Sep 2011

Certificateless Cryptography With Kgc Trust Level 3, Guomin Yang, Chik How Tan

Research Collection School Of Computing and Information Systems

A normal certificateless cryptosystem can only achieve KGC trust level 2 according to the trust hierarchy defined by Girault. Although in the seminal paper introducing certificateless cryptography, Al-Riyami and Paterson introduced a binding technique to lift the KGC trust level of their certificateless schemes to level 3, many subsequent work on certificateless cryptography just focused on the constructions of normal certificateless schemes, and a formal study on the general applicability of the binding technique to these existing schemes is still missing. In this paper, to address the KGC trust level issue, we introduce the notion of Key Dependent Certificateless Cryptography …

Go to article

Privacy Beyond Single Sensitive Attribute, Yuan Fang, Mafruz Zaman Ashrafi, See Kiong Ng Sep 2011

Privacy Beyond Single Sensitive Attribute, Yuan Fang, Mafruz Zaman Ashrafi, See Kiong Ng

Research Collection School Of Computing and Information Systems

Publishing individual specific microdata has serious privacy implications. The k-anonymity model has been proposed to prevent identity disclosure from microdata, and the work on ℓ-diversity and t-closeness attempt to address attribute disclosure. However, most current work only deal with publishing microdata with a single sensitive attribute (SA), whereas real life scenarios often involve microdata with multiple SAs that may be multi-valued. This paper explores the issue of attribute disclosure in such scenarios. We propose a method called CODIP (Complete Disjoint Projections) that outlines a general solution to deal with the shortcomings in a naïve approach. We also introduce two measures, …

Go to article

Towards Ground Truthing Observations In Gray-Box Anomaly Detection, Jiang Ming, Haibin Zhang, Debin Gao Sep 2011

Towards Ground Truthing Observations In Gray-Box Anomaly Detection, Jiang Ming, Haibin Zhang, Debin Gao

Research Collection School Of Computing and Information Systems

Anomaly detection has been attracting interests from researchers due to its advantage of being able to detect zero-day exploits. A gray-box anomaly detector first observes benign executions of a computer program and then extracts reliable rules that govern the normal execution of the program. However, such observations from benign executions are not necessarily true evidences supporting the rules learned. For example, the observation that a file descriptor being equal to a socket descriptor should not be considered supporting a rule governing the two values to be the same. Ground truthing such observations is a difficult problem since it is not …

Go to article

Improved Ordinary Measure And Image Entropy Theory Based Intelligent Copy Detection Method, Dengpan Ye, Longfei Ma, Lina Wang, Robert H. Deng Sep 2011

Improved Ordinary Measure And Image Entropy Theory Based Intelligent Copy Detection Method, Dengpan Ye, Longfei Ma, Lina Wang, Robert H. Deng

Research Collection School Of Computing and Information Systems

Nowadays, more and more multimedia websites appear in social network. It brings some security problems, such as privacy, piracy, disclosure of sensitive contents and so on. Aiming at copyright protection, the copy detection technology of multimedia contents becomes a hot topic. In our previous work, a new computer-based copyright control system used to detect the media has been proposed. Based on this system, this paper proposes an improved media feature matching measure and an entropy based copy detection method. The Levenshtein Distance was used to enhance the matching degree when using for feature matching measure in copy detection. For entropy …

Go to article

Defending Against Cross Site Scripting Attacks, Lwin Khin Shar, Hee Beng Kuan Tan Aug 2011

Defending Against Cross Site Scripting Attacks, Lwin Khin Shar, Hee Beng Kuan Tan

Research Collection School Of Computing and Information Systems

Researchers have proposed multiple solutions to cross-site scripting, but vulnerabilities continue to exist in many Web applications due to developers' lack of understanding of the problem and their unfamiliarity with current defenses' strengths and limitations.

Go to article

A Hubel Wiesel Model Of Early Concept Generalization Based On Local Correlation Of Input Features, Sepideh Sadeghi, Kiruthika Ramanathan Aug 2011

A Hubel Wiesel Model Of Early Concept Generalization Based On Local Correlation Of Input Features, Sepideh Sadeghi, Kiruthika Ramanathan

Research Collection School Of Computing and Information Systems

Hubel Wiesel models, successful in visual processing algorithms, have only recently been used in conceptual representation. Despite the biological plausibility of a Hubel-Wiesel like architecture for conceptual memory and encouraging preliminary results, there is no implementation of how inputs at each layer of the hierarchy should be integrated for processing by a given module, based on the correlation of the features. In our paper, we propose the input integration framework - a set of operations performed on the inputs to the learning modules of the Hubel Wiesel model of conceptual memory. These operations weight the modules as being general or …

Go to article

Taxisim: A Multiagent Simulation Platform For Evaluating Taxi Fleet Operations, Shih-Fen Cheng, Thi Duong Nguyen Aug 2011

Taxisim: A Multiagent Simulation Platform For Evaluating Taxi Fleet Operations, Shih-Fen Cheng, Thi Duong Nguyen

Research Collection School Of Computing and Information Systems

Taxi service is an important mode of public transportation in most metropolitan areas since it provides door-to-door convenience in the public domain. Unfortunately, despite all the convenience taxis bring, taxi fleets are also extremely inefficient to the point that over 50% of its operation time could be spent in idling state. Improving taxi fleet operation is an extremely challenging problem, not just because of its scale, but also due to fact that taxi drivers are self-interested agents that cannot be controlled centrally. To facilitate the study of such complex and decentralized system, we propose to construct a multiagent simulation platform …

Go to article

A Generic Framework For Three-Factor Authentication: Preserving Security And Privacy In Distributed Systems, Xinyi Huang, Yang Xiang, Ashley Chonka, Jianying Zhou, Robert H. Deng Aug 2011

A Generic Framework For Three-Factor Authentication: Preserving Security And Privacy In Distributed Systems, Xinyi Huang, Yang Xiang, Ashley Chonka, Jianying Zhou, Robert H. Deng

Research Collection School Of Computing and Information Systems

As part of the security within distributed systems, various services and resources need protection from unauthorized use. Remote authentication is the most commonly used method to determine the identity of a remote client. This paper investigates a systematic approach for authenticating clients by three factors, namely password, smart card, and biometrics. A generic and secure framework is proposed to upgrade two-factor authentication to three-factor authentication. The conversion not only significantly improves the information assurance at low cost but also protects client privacy in distributed systems. In addition, our framework retains several practice-friendly properties of the underlying two-factor authentication, which we …

Go to article

General Construction Of Chameleon All-But-One Trapdoor Functions, Shengli Liu, Junzuo Lai, Robert H. Deng Aug 2011

General Construction Of Chameleon All-But-One Trapdoor Functions, Shengli Liu, Junzuo Lai, Robert H. Deng

Research Collection School Of Computing and Information Systems

Lossy trapdoor functions enable black-box construction of public key encryption (PKE) schemes secure against chosen-ciphertext attack [18]. Recently, a more efficient black-box construction of public key encryption was given in [13] with the help of chameleon all-but-one trapdoor functions (ABO-TDFs). In this paper, we propose a black-box construction for transforming any ABO-TDFs into chameleon ABO-TDFs with the help of chameleon hash functions. Instantiating the proposed general black-box construction of chameleon ABO-TDFs, we obtain the first chameleon ABO-TDFs based on the Decisional Diffie-Hellman (DDH) assumption.

Go to article

Ccrank: Parallel Learning To Rank With Cooperative Coevolution, Shuaiqiang Wang, Byron J. Gao, Ke Wang, Hady W. Lauw Aug 2011

Ccrank: Parallel Learning To Rank With Cooperative Coevolution, Shuaiqiang Wang, Byron J. Gao, Ke Wang, Hady W. Lauw

Research Collection School Of Computing and Information Systems

We propose CCRank, the first parallel algorithm for learning to rank, targeting simultaneous improvement in learning accuracy and efficiency. CCRank is based on cooperative coevolution (CC), a divide-and-conquer framework that has demonstrated high promise in function optimization for problems with large search space and complex structures. Moreover, CC naturally allows parallelization of sub-solutions to the decomposed subproblems, which can substantially boost learning efficiency. With CCRank, we investigate parallel CC in the context of learning to rank. Extensive experiments on benchmarks in comparison with the state-of-the-art algorithms show that CCRank gains in both accuracy and efficiency.

Go to article

Parallel Learning To Rank For Information Retrieval, Shuaiqiang Wang, Byron J. Gao, Ke Wang, Hady W. Lauw Jul 2011

Parallel Learning To Rank For Information Retrieval, Shuaiqiang Wang, Byron J. Gao, Ke Wang, Hady W. Lauw

Research Collection School Of Computing and Information Systems

Learning to rank represents a category of effective ranking methods for information retrieval. While the primary concern of existing research has been accuracy, learning efficiency is becoming an important issue due to the unprecedented availability of large-scale training data and the need for continuous update of ranking functions. In this paper, we investigate parallel learning to rank, targeting simultaneous improvement in accuracy and efficiency.

Go to article

Automated Detection Of Likely Design Flaws In Layered Architectures, Aditya Budi, - Lucia, David Lo, Lingxiao Jiang, Shaowei Wang Jul 2011

Automated Detection Of Likely Design Flaws In Layered Architectures, Aditya Budi, - Lucia, David Lo, Lingxiao Jiang, Shaowei Wang

Research Collection School Of Computing and Information Systems

Layered architecture prescribes a good principle for separating concerns to make systems more maintainable. One example of such layered architectures is the separation of classes into three groups: Boundary, Control, and Entity, which are referred to as the three analysis class stereotypes in UML. Classes of different stereotypes are interacting with one another, when properly designed, the overall interaction would be maintainable, flexible, and robust. On the other hand, poor design would result in less maintainable system that is prone to errors. In many software projects, the stereotypes of classes are often missing, thus detection of design flaws becomes non-trivial. …

Go to article

Real-World Parameter Tuning Using Factorial Design With Parameter Decomposition, Aldy Gunawan, Hoong Chuin Lau, Elaine Wong Jul 2011

Real-World Parameter Tuning Using Factorial Design With Parameter Decomposition, Aldy Gunawan, Hoong Chuin Lau, Elaine Wong

Research Collection School Of Computing and Information Systems

In this paper, we explore the idea of improving the efficiency of factorial design for parameter tuning of metaheuristics. In a standard full factorial design, the number of runs increases exponentially as the number of parameters. To reduce the parameter search space, one option is to first partition parameters into disjoint categories. While this may be done manually based on user guidance, an automated approach proposed in this paper is to apply a fractional factorial design to partition parameters based on their main effects where each partition is then tuned independently. With a careful choice of fractional design, our approach …

Go to article

Unsupervised Discovery Of Discourse Relations For Eliminating Intra-Sentence Polarity Ambiguities, Lanjun Zhou, Binyang Li, Wei Gao, Zhongyu Wei, Kam-Fai Wong Jul 2011

Unsupervised Discovery Of Discourse Relations For Eliminating Intra-Sentence Polarity Ambiguities, Lanjun Zhou, Binyang Li, Wei Gao, Zhongyu Wei, Kam-Fai Wong

Research Collection School Of Computing and Information Systems

Polarity classification of opinionated sentences with both positive and negative sentiments1 is a key challenge in sentiment analysis. This paper presents a novel unsupervised method for discovering intra-sentence level discourse relations for eliminating polarity ambiguities. Firstly, a discourse scheme with discourse constraints on polarity was defined empirically based on Rhetorical Structure Theory (RST). Then, a small set of cuephrase-based patterns were utilized to collect a large number of discourse instances which were later converted to semantic sequential representations (SSRs). Finally, an unsupervised method was adopted to generate, weigh and filter new SSRs without cue phrases for recognizing discourse relations. Experimental …

Go to article

Relevant Knowledge Helps In Choosing Right Teacher: Active Query Selection For Ranking Adaptation, Peng Cai, Wei Gao, Kam-Fai Wong, Aoying Zhou Jul 2011

Relevant Knowledge Helps In Choosing Right Teacher: Active Query Selection For Ranking Adaptation, Peng Cai, Wei Gao, Kam-Fai Wong, Aoying Zhou

Research Collection School Of Computing and Information Systems

Learning to adapt in a new setting is a common challenge to our knowledge and capability. New life would be easier if we actively pursued supervision from the right mentor chosen with our relevant but limited prior knowledge. This variant principle of active learning seems intuitively useful to many domain adaptation problems. In this paper, we substantiate its power for advancing automatic ranking adaptation, which is important in web search since it's prohibitive to gather enough labeled data for every search domain for fully training domain-specific rankers. For the cost-effectiveness, it is expected that only those most informative instances in …

Go to article

Generating Aspect-Oriented Multi-Document Summarization With Event-Aspect Model, Peng Li, Yinglin Wang, Wei Gao, Jing Jiang Jul 2011

Generating Aspect-Oriented Multi-Document Summarization With Event-Aspect Model, Peng Li, Yinglin Wang, Wei Gao, Jing Jiang

Research Collection School Of Computing and Information Systems

In this paper, we propose a novel approach to automatic generation of aspect-oriented summaries from multiple documents. We first develop an event-aspect LDA model to cluster sentences into aspects. We then use extended LexRank algorithm to rank the sentences in each cluster. We use Integer Linear Programming for sentence selection. Key features of our method include automatic grouping of semantically related sentences and sentence ranking based on extension of random walk model. Also, we implement a new sentence compression algorithm which use dependency tree instead of parser tree. We compare our method with four baseline methods. Quantitative evaluation based on …

Go to article

Scalable Multiagent Planning Using Probabilistic Inference, Akshat Kumar, Shlomo Zilberstein, Marc Toussaint Jul 2011

Scalable Multiagent Planning Using Probabilistic Inference, Akshat Kumar, Shlomo Zilberstein, Marc Toussaint

Research Collection School Of Computing and Information Systems

Multiagent planning has seen much progress with the development of formal models such as Dec-POMDPs. However, the complexity of these models -- NEXP-Complete even for two agents -- has limited scalability. We identify certain mild conditions that are sufficient to make multiagent planning amenable to a scalable approximation w.r.t. the number of agents. This is achieved by constructing a graphical model in which likelihood maximization is equivalent to plan optimization. Using the Expectation-Maximization framework for likelihood maximization, we show that the necessary inference can be decomposed into processes that often involve a small subset of agents, thereby facilitating scalability. We …

Go to article

Online Auc Maximization, Peilin Zhao, Steven C. H. Hoi, Rong Jin, Tianbo Yang Jul 2011

Online Auc Maximization, Peilin Zhao, Steven C. H. Hoi, Rong Jin, Tianbo Yang

Research Collection School Of Computing and Information Systems

Most studies of online learning measure the performance of a learner by classification accuracy, which is inappropriate for applications where the data are unevenly distributed among different classes. We address this limitation by developing online learning algorithm for maximizing Area Under the ROC curve (AUC), a metric that is widely used for measuring the classification performance for imbalanced data distributions. The key challenge of online AUC maximization is that it needs to optimize the pairwise loss between two instances from different classes. This is in contrast to the classical setup of online learning where the overall loss is a sum …

Go to article

First Prev Next Last