Title

1

Presenter: Chen Manman (chenman@comp.nus.edu.sg)


Title:
Ensuring Security, Reliability and Robustness of Web Application

Abstract:

Web and cloud applications nowadays are ubiquitous. Nevertheless, any vulnerability in these applications is susceptible to be exploited by malicious users, which form serious security problems. In addition, the reliability and robustness of the applications are difficult to guarantee, due to the unpredictable nature of the internet. To this end, we have conducted a series of works that aim to efficiently and effectively improve the security, reliability and robustness of Web application. Some of the works that have been conducted are shown in the following.

1. Web Model Checker – We have developed a Web model checker, VeriWS, for verifying of Web application on functional and non-functional correctness. With a push of button, the user can verify the functional and non-functional correctness for their Web applications automatically.

2. Malware Analyser - We have developed a malware checker for Web applications based on machine learning techniques, which has shown to outperform all existing off-the-shelf commercial anti-viruses.

3. Automatic Recovery of Web Application - We have devised a method that allows automatic recovery of Web application during the failure of the application. This effectively and significantly minimizes or even avoids the cost that brought by the failure.

4. Requirements Synthesis for Web Application - We have proposed a methodology in synthesizing time requirements for components for Web application. The synthesized time requirements allow automatic selection of components that satisfy the requirements.


2

Presenter: Guo Long (guolong@comp.nus.edu.sg)

Title:
Location-Aware Pub/Sub System: When Continuous Moving Queries Meet Dynamic Event Streams

Abstract:
In this paper, we propose a new location-aware pub/sub system, Elaps, that continuously monitors moving users subscribing to dynamic event streams from social media and E-commerce applications. Users are notified instantly when there is a matching event nearby. To the best of our knowledge, Elaps is the first to take into account continuous moving queries against dynamic event streams. Like existing works on continuous moving query processing, Elaps employs the concept of safe region to reduce communication overhead. However, unlike existing works which assume data from publishers are static, updates to safe regions may be triggered by newly arrived events. In Elaps, we develop a concept called impact region that allows us to identify whether a safe region is affected by newly arrived events. Moreover, we propose a novel cost model to optimize the safe region size to keep the communication overhead low. Based on the cost model, we design two incremental methods, iGM and idGM, for safe region construction. In addition, Elaps uses boolean expression, which is more expressive than keywords, to model user intent and we propose a novel index, BEQ-Tree, to handle spatial boolean expression matching. In our experiments, we use geo-tweets from Twitter and venues from Foursquare to simulate publishers and boolean expressions generated from AOL search log to represent users intentions. We test user movement in both synthetic trajectories and real taxi trajectories. The results show that Elaps can significantly reduce the communication overhead and disseminate events to users in real-time.


3

Presenter:
Kartik Sankaran (kartiks@comp.nus.edu.sg)

Title: Using Mobile Phone Barometer for Low-Power Transportation Context Detection

Abstract:
We propose and implement a system that uses only the smartphone's barometer sensor for low-power context detection of the states idle, walking, and vehicle. Unlike existing approaches that have high user dependence and require extensive training, our barometer-based approach is inherently independent of the user, having similar detection accuracy while consuming lower power. In particular, when the user is waiting for a bus, and while traveling on smooth vehicles, we find that Google's accelerometer-based activity recognition algorithm has less than 25% detection accuracy, while our approach has almost 100% accuracy.



4

Presenter:
Kartik Sankaran (kartiks@comp.nus.edu.sg)

Title:
Smartphone-based Bus Route and Bus-stop Detection using Barometer and Phone-to-Phone Communication

Abstract:
We propose novel techniques for bus route and bus-stop detection using only the smartphone's barometer sensor and phone-to-phone communication between users in the same bus. Our approach makes the fewest assumptions on infrastructure and availability of route maps, enabling it to work in even in developing countries where other approaches fail. It is decentralized, does not require an Internet connection, and additionally provides location privacy while reducing power consumption.


5

Presenter: Li Yuchen (a0047194@comp.nus.edu.sg)

Title:
Extreme Search on Social Networks

Abstract:
Online Social Networks (OSN) data is large, fast and diverse. First, the size of OSN data is huge. For example, there are 1.44 billion active users on Facebook in 2015. Second, the rate at which social networks generate information is tremendous. Twitter has over 500 million tweets being sent per day. Lastly, OSN data is diverse. Today’s social networks are embedded with heterogeneous information including textual, spatial and multimedia data. Such huge and rich data often overwhelms social network users. Keyword search is an important tool to facilitate users in consuming these data. The goal of my research is to deliver the right information to the right person at the right time and the right place on OSN. However, such goal is exciting and, at the same time, very challenging.  Besides the above-mentioned unique properties of OSN data, we need to deal with the underlying complex social graphs, if we wish to achieve this goal. There are many intrinsic characteristics of OSN like power law, small world, sparse and high clustering property. Therefore, my research contributes to this endeavour by focusing on leveraging the unique properties of OSN to deliver effective keyword search solutions to OSN users.

6

Presenter: Lim Zhan Wei (
limzhanw@comp.nus.edu.sg)

Title:
Adaptive Stochastic Optimization: From Sets to Paths

Abstract:
Adaptive stochastic optimization (ASO) optimizes an objective function adaptively under uncertainty. It plays a crucial role in planning and learning under uncertainty, but is, unfortunately, computationally intractable in general.  We introduce two conditions on the objective function, the marginal likelihood rate bound and the marginal likelihood bound, which, together with pointwise submodularity, enable efficient approximate solution of ASO. Several interesting classes of functions satisfy these conditions naturally, e.g., the version space reduction function for hypothesis learning.  We describe Recursive Adaptive Coverage, a new ASO algorithm that exploits these conditions, and apply the algorithm to two robot planning tasks under uncertainty. In contrast to the earlier submodular optimization approach, our algorithm applies to ASO over both sets and paths.


7

Presenter: Pham Van Thuan (thuanpv@comp.nus.edu.sg)

Title:
Title: Hercules: Reproducing Crashes in Real-World Application Binaries

Abstract:
Binary analysis is a well-investigated area in software engineering and security. Given real-world program binaries, generating test inputs which cause the binaries to crash is crucial. Generation of crashing inputs has many applications including off-line analysis of software prior to deployment, or online analysis of software patches as they are inserted. In this work, we present a method for generating inputs which reach a given "potentially crashing" location. Such potentially crashing locations can be found by a separate static analysis (or by gleaning crash reports submitted by internal / external users) and serve as the input to our method. The test input generated by our method serves as a witness of the crash. Our method is particularly suited for binaries of programs which take in complex structured inputs. Experiments on real-life applications such as the Adobe Reader and the Windows Media Player demonstrate that our Hercules tool built on selective symbolic execution engine S2E can generate crashing inputs within few hours, where symbolic approaches (as embodied by S2E) or blackbox fuzzing approaches (as embodied by the commercial tool PeachFuzzer) failed.

8

Presenter: Song Xuemeng (xuemeng@comp.nus.edu.sg)

Title:
Interest Inference via Structure-Constrained Multi-Source Multi-Task Learning

Abstract:
User interest inference from social networks is a fundamental problem to many applications. It usually exhibits dual-heterogeneities: a user’s interests are complementarily and comprehensively reflected by multiple social networks; interests are inter-correlated in a non-uniform way rather than independent to each other. Although great success has been achieved by previous approaches, few of them consider these dual-heterogeneities simultaneously. In this work, we propose a structure-constrained multi-source multi-task learning scheme to co-regularize the source consistency and the tree-guided task relatedness. Meanwhile, it is able to jointly learn the task sharing and task-specific features. Comprehensive experiments on a real-world dataset validated our scheme. In addition, we have released our dataset to facilitate the research communities.

9

Presenter: Tan Shin Hwei (shinhwei@comp.nus.edu.sg)

Title:
Relifix: Automated Repair of Software Regressions

Abstract:
Regression occurs when code changes introduce failures in previously passing test cases. As software evolves, regressions may be introduced. Fixing regression errors manually is time-consuming and error-prone. We propose an approach of automated repair of software regressions, called relifix, which considers the regression repair problem as a problem of reconciling problematic changes. Specifically, we derive a set of code transformations obtained from our manual inspection of 73 real software regressions; this set of code transformations uses syntactical information from changed statements. Regression repair is then accomplished via a search over the code transformation operators – which operator to apply, and where. Our evaluation compares the repairability of relifix with GenProg on 35 real regression errors. relifix repairs 23 bugs, while GenProg only fixes five bugs. We also measure the likelihood of both approaches in introducing new regressions given a reduced test suite. Our experimental results show that our approach is less likely to introduce new regressions than GenProg.


10

Presenter: W.K.P.N.N.R Goonawardene (ngoonawa@comp.nus.edu.sg)

Title:
Do online communities help patients to achieve health goals? The role of sub-group cultures and progression spiral effects

Abstract:
With the popularity of Health 2.0, healthcare social networks have become powerful tools of bringing people with shared health interests together. These platforms are progressively implementing functionalities such as quantified self-tracking and collaborative filtering to identify potentially related conditions of patients and match patients in similar situations, enabling patients to collectively conduct healthcare behaviours in virtual environments.  Participation in online healthcare networks is potentially useful for patients to internalize healthy behaviours. Up to date, only a little research has been done to understand the influence of online communities on patients’ self-regulatory healthcare behaviours. In this research we study the influence of online healthcare networks on achieving self-regulatory healthcare targets of patients. We obtained and analysed data based on patient-generated content and user profiles of a popular online healthcare community.



11

Presenter: Zheng Chaodong (zheng-10@comp.nus.edu.sg)

Title:
Efficient Communication in Cognitive Radio Networks

Abstract:
Devices in a cognitive radio network use advanced radios to identify pockets of usable spectrum in a crowded band and make them available to higher layers of the network stack. A core challenge in designing algorithms for this model is that different devices might have different views of the network. Recently, we have studied two problems for this setting that are well-motivated but not yet well-understood: local broadcast and data aggregation. In particular, we consider a single hop cognitive radio network with n nodes that each has access to c channels. We assume each pair of nodes overlaps on at least 1 <= k <= c channels. We first describe and analyse CogCast, a randomized algorithm that solves local broadcast efficiently, by spreading information in an epidemic manner through the network. We then propose CogComp, a randomized algorithm that solves data aggregation efficiently. The CogComp algorithm uses CogCast as a key primitive to establish a spanning tree among the nodes, so that data can be aggregated from leaves to root. We have also developed a collection of lower bounds that show CogCast is near optimal in many cases. These bounds introduce new techniques of potential standalone interest for those concerned with proving fundamental limits in the cognitive radio network setting. Our next step in this research is to extend the results to the multi-hop scenario.