Prosody-Based Query-By-Example Spoken Utterance Retrieval,
2012
Department of Computer Science, University of Texas at El Paso
Prosody-Based Query-By-Example Spoken Utterance Retrieval, Steven D. Werner, Nigel G. Ward
COURI Symposium Abstracts, Summer 2012
Current spoken dialogue retrieval systems are almost entirely lexically based, using automatic transcription by imperfect speech recognition technology. Once audio is reduced to text, little more than existing text search methods are used to retrieve results. However, spoken dialogue contains rich latent prosodic information about dialogue activities - such as agreeing, arguing, deciding, planning, storytelling, showing surprise - and this extra information can often be relevant to the searchers aims. Searching for similar speech acts rather then similar words enables similar contextual results, returning points where the speaker had the same sentiment during a topic. Previous research yielded 76 dimensions …
The Latent Maximum Entropy Principle,
2012
Wright State University - Main Campus
The Latent Maximum Entropy Principle, Shaojun Wang, Dale Schuurmans, Yunxin Zhao
Kno.e.sis Publications
We present an extension to Jaynes’ maximum entropy principle that incorporates latent variables. The principle of latent maximum entropy we propose is different from both Jaynes’ maximum entropy principle and maximum likelihood estimation, but can yield better estimates in the presence of hidden variables and limited training data. We first show that solving for a latent maximum entropy model poses a hard nonlinear constrained optimization problem in general. However, we then show that feasible solutions to this problem can be obtained efficiently for the special case of log-linear models---which forms the basis for an efficient approximation to the latent maximum …
What Kind Of #Communication Is Twitter? A Psycholinguistic Perspective On Communication In Twitter For The Purpose Of Emergency Coordination,
2012
Wright State University - Main Campus
What Kind Of #Communication Is Twitter? A Psycholinguistic Perspective On Communication In Twitter For The Purpose Of Emergency Coordination, Hemant Purohit, Andrew Hampton, Valerie L. Shalin, Amit P. Sheth, John Flach
Kno.e.sis Publications
The present research aims to detect coordinated citizen response within social media traffic to assist emergency response. We use domain-independent linguistic properties as the first step in narrowing the candidate set of messages for domain-dependent and computationally intensive analysis.
Exploring The Impact Of Knowledge And Social Environment On Influenza Prevention And Transmission In Midwestern United States High School Students,
2012
Wright State University - Main Campus
Exploring The Impact Of Knowledge And Social Environment On Influenza Prevention And Transmission In Midwestern United States High School Students, William L. Romine, Tanvi Banerjee, William S. Barrow, William R. Folk
Kno.e.sis Publications
We used data from a convenience sample of 410 Midwestern United States students from six secondary schools to develop parsimonious models for explaining and predicting precautions and illness related to influenza. Scores for knowledge and perceptions were obtained using two-parameter Item Response Theory (IRT) models. Relationships between outcome variables and predictors were verified using Pearson and Spearman correlations, and nested [student within school] fixed effects multinomial logistic regression models were specified from these using Akaike’s Information Criterion (AIC). Neural network models were then formulated as classifiers using 10-fold cross validation to predict precautions and illness. Perceived barriers against taking precautions …
Determinants In Sustaining A Local Information System In The Philippines: The Case Of The Barangay Management Information System (Bmis),
2012
University of the Philippines Los Baños
Determinants In Sustaining A Local Information System In The Philippines: The Case Of The Barangay Management Information System (Bmis), Charina P. Maneja, Nancy A. Tandang, Merlyne M. Paunlagui
Journal of Public Affairs and Development
Information is important for the executive and legislative functions of local officials. The study determined the institutional and individual factors that contributed in sustaining a Barangay Management Information System (BMIS). The study was done in five provinces covering 90 randomly selected continuing barangays and 68 randomly selected non-continuing barangays. Chi-square Test of Independence was used to determine factors associated with whether the barangay will continue to sustain BMIS or not. Logistic regression analysis was also performed to determine factors that may influence barangay's decision to sustain BMIS. The identified significant individual factors that influenced the barangays' decision to sustain BMIS …
K-Partite Graph Reinforcement And Its Application In Multimedia Information Retrieval,
2012
Huazhong University of Science and Technology
K-Partite Graph Reinforcement And Its Application In Multimedia Information Retrieval, Yue Gao, Meng Wang, Rongrong Ji, Zheng-Jun Zha, Jialie Shen
Research Collection School Of Computing and Information Systems
In many example-based information retrieval tasks, example query actually contains multiple sub-queries. For example, in 3D object retrieval, the query is an object described by multiple views. In content-based video retrieval, the query is a video clip that contains multiple frames. Without prior knowledge, the most intuitive approach is to treat the sub-queries equally without difference. In this paper, we propose a k-partite graph reinforcement approach to fuse these sub-queries based on the to-be-retrieved database. The approach first collects the top retrieved results. These results are regarded as pseudo-relevant samples and then a k-partite graph reinforcement is performed on these …
Identifying Event-Related Bursts Via Social Media Activities,
2012
Singapore Management University
Identifying Event-Related Bursts Via Social Media Activities, Xin Zhao, Baihan Shu, Jing Jiang, Yang Song, Hongfei Yan, Xiaoming Li
Research Collection School Of Computing and Information Systems
Activities on social media increase at a dramatic rate. When an external event happens, there is a surge in the degree of activities related to the event. These activities may be temporally correlated with one another, but they may also capture different aspects of an event and therefore exhibit different bursty patterns. In this paper, we propose to identify event-related bursts via social media activities. We study how to correlate multiple types of activities to derive a global bursty pattern. To model smoothness of one state sequence, we propose a novel function which can capture the state context. The experiments …
Exact Soft Confidence-Weighted Learning,
2012
Nanyang Technological University
Exact Soft Confidence-Weighted Learning, Jialei Wang, Steven C. H. Hoi
Research Collection School Of Computing and Information Systems
In this paper, we propose a new Soft Confidence-Weighted (SCW) online learning scheme, which enables the conventional confidence-weighted learning method to handle non-separable cases. Unlike the previous confidence-weighted learning algorithms, the proposed soft confidence-weighted learning method enjoys all the four salient properties: (i) large margin training, (ii) confidence weighting, (iii) capability to handle non-separable data, and (iv) adaptive margin. Our experimental results show that the proposed SCW algorithms significantly outperform the original CW algorithm. When comparing with a variety of state-of-the art algorithms (including AROW, NAROW and NHERD), we found that SCW generally achieves better or at least comparable predictive …
Online Kernel Selection: Algorithms And Evaluations,
2012
Michigan State University
Online Kernel Selection: Algorithms And Evaluations, Tianbao Yang, Mehrdad Mahdavi, Rong Jin, Jinfeng Yi, Steven C. H. Hoi
Research Collection School Of Computing and Information Systems
Kernel methods have been successfully applied to many machine learning problems. Nevertheless, since the performance of kernel methods depends heavily on the type of kernels being used, identifying good kernels among a set of given kernels is important to the success of kernel methods. A straightforward approach to address this problem is cross-validation by training a separate classifier for each kernel and choosing the best kernel classifier out of them. Another approach is Multiple Kernel Learning (MKL), which aims to learn a single kernel classifier from an optimal combination of multiple kernels. However, both approaches suffer from a high computational …
Mydeal: The Context-Aware Urban Shopping Assistant,
2012
Singapore Management University
Mydeal: The Context-Aware Urban Shopping Assistant, Kartik Muralidharan, Swapna Gottipati, Jing Jiang, Narayan Ramasubbu, Rajesh Krishna Balan
Research Collection School Of Computing and Information Systems
A common problem in large Urban cities, of the sort seen in Asia, is the huge number of retail options available in the city. In particular, it is not uncommon to find multiple malls, each with hundreds of stores inside, just a short distance from each other in almost every part of these cities. These factors make it incredibly hard for consumers to identify stores of interest to them in any particular mall.In response, a number of shopping assistance applications have been created for mobile phones.However, these applications mostly just allow users to know which stores are where or to …
Finding Bursty Topics From Microblogs,
2012
Singapore Management University
Finding Bursty Topics From Microblogs, Qiming Diao, Jing Jiang, Feida Zhu, Ee Peng Lim
Research Collection School Of Computing and Information Systems
Microblogs such as Twitter reflect the general public’s reactions to major events. Bursty topics from microblogs reveal what events have attracted the most online attention. Although bursty event detection from text streams has been studied before, previous work may not be suitable for microblogs because compared with other text streams such as news articles and scientific publications, microblog posts are particularly diverse and noisy. To find topics that have bursty patterns on microblogs, we propose a topic model that simultaneousy captures two observations: (1) posts published around the same time are more likely to have the same topic, and (2) …
Adaptive Cgf For Pilots Training In Air Combat Simulation,
2012
Singapore Management University
Adaptive Cgf For Pilots Training In Air Combat Simulation, Teck-Hou Teng, Ah-Hwee Tan, Wee-Sze Ong, Kien-Lip Lee
Research Collection School Of Computing and Information Systems
Training of combat fighter pilots is often conducted using either human opponents or non-adaptive computer-generated force (CGF) inserted with the doctrine for conducting air combat mission. The novelty and challenges of such non-adaptive doctrine-driven CGF is often lost quickly. Incorporating more complex knowledge manually is known to be tedious and time-consuming. Therefore, a study of using adaptive CGF to learn from the real-time interactions with human pilots to extend the existing doctrine is conducted in this work. The goal of this study is to show how an adaptive CGF can be more effective than a non-adaptive doctrine-driven CGF for simulator-based …
On-Line Portfolio Selection With Moving Average Reversion,
2012
Nanyang Technological University
On-Line Portfolio Selection With Moving Average Reversion, Bin Li, Steven C. H. Hoi
Research Collection School Of Computing and Information Systems
On-line portfolio selection has attracted increasing interests in machine learning and AI communities recently. Empirical evidences show that stock's high and low prices are temporary and stock price relatives are likely to follow the mean reversion phenomenon. While the existing mean reversion strategies are shown to achieve good empirical performance on many real datasets, they often make the single-period mean reversion assumption, which is not always satisfied in some real datasets, leading to poor performance when the assumption does not hold. To overcome the limitation, this article proposes a multiple-period mean reversion, or so-called Moving Average Reversion (MAR), and a …
Fast Bounded Online Gradient Descent Algorithms For Scalable Kernel-Based Online Learning,
2012
Nanyang Technological University
Fast Bounded Online Gradient Descent Algorithms For Scalable Kernel-Based Online Learning, Peilin Zhao, Jialei Wang, Pengcheng Wu, Rong Jin, Steven C. H. Hoi
Research Collection School Of Computing and Information Systems
Kernel-based online learning has often shown state-of-the-art performance for many online learning tasks. It, however, suffers from a major shortcoming, that is, the unbounded number of support vectors, making it non-scalable and unsuitable for applications with large-scale datasets. In this work, we study the problem of bounded kernel-based online learning that aims to constrain the number of support vectors by a predefined budget. Although several algorithms have been proposed in literature, they are neither computationally efficient due to their intensive budget maintenance strategy nor effective due to the use of simple Perceptron algorithm. To overcome these limitations, we propose a …
Topic Discovery From Tweet Replies,
2012
Singapore Management University
Topic Discovery From Tweet Replies, Bingtian Dai, Ee Peng Lim, Philips Kokoh Prasetyo
Research Collection School Of Computing and Information Systems
Twitter is a popular online social information network service which allows people to read and post messages up to 140 characters, known as “tweets”. In this paper, we focus on the tweets between pairs of individuals, i.e., the tweet replies, and propose a generative model to discover topics among groups of twitter users. Our model has then been evaluated with a tweet dataset to show its effectiveness.
Enhancing Access Privacy Of Range Retrievals Over B+Trees,
2012
Singapore Management University
Enhancing Access Privacy Of Range Retrievals Over B+Trees, Hwee Hwa Pang, Jilian Zhang, Kyriakos Mouratidis
Research Collection School Of Computing and Information Systems
Users of databases that are hosted on shared servers cannot take for granted that their queries will not be disclosed to unauthorized parties. Even if the database is encrypted, an adversary who is monitoring the I/O activity on the server may still be able to infer some information about a user query. For the particular case of a B+-tree that has its nodes encrypted, we identify properties that enable the ordering among the leaf nodes to be deduced. These properties allow us to construct adversarial algorithms to recover the B+-tree structure from the I/O traces generated by range queries. Combining …
Joint Learning For Coreference Resolution With Markov Logic,
2012
Peking University
Joint Learning For Coreference Resolution With Markov Logic, Yang Song, Jing Jiang, Xin Zhao, Sujian Li, Houfeng Wang
Research Collection School Of Computing and Information Systems
Pairwise coreference resolution models must merge pairwise coreference decisions to generate final outputs. Traditional merging methods adopt different strategies such as the best first method and enforcing the transitivity constraint, but most of these methods are used independently of the pairwise learning methods as an isolated inference procedure at the end. We propose a joint learning model which combines pairwise classification and mention clustering with Markov logic. Experimental results show that our joint learning system outperforms independent learning systems. Our system gives a better performance than all the learning-based systems from the CoNLL-2011 shared task on the same dataset. Compared …
Detecting Anomalous Twitter Users By Extreme Group Behaviors,
2012
Singapore Management University
Detecting Anomalous Twitter Users By Extreme Group Behaviors, Hanbo Dai, Ee-Peng Lim, Feida Zhu, Hwee Hwa Pang
Research Collection School Of Computing and Information Systems
Twitter has enjoyed tremendous popularity in the recent years. To help categorizing and search tweets, Twitter users assign hashtags to their tweets. Given that hashtag assignment is the primary way to semantically categorizing and search tweets, it is highly susceptible to abuse by spammers and other anomalous users [1]. Popular hashtags such as #Obama and #ladygaga could be hijacked by having them added to unrelated tweets with the intent of misleading many other users or promoting specific agenda to the users. The users performing this act are known as the hashtag hijackers. As the hijackers usually abuse common sets of …
Formal Analysis Of Pervasive Computing Systems,
2012
Singapore Management University
Formal Analysis Of Pervasive Computing Systems, Yan Liu, Xian Zhang, Jin Song Dong, Yang Liu, Jun Sun, Jit Biswas, Mounir Mokhtari
Research Collection School Of Computing and Information Systems
Pervasive computing systems are heterogenous and complex as they usually involve human activities, various sensors and actuators as well as middleware for system controlling. Therefore, analyzing such systems is highly nontrivial. In this work, we propose to use formal methods for analyzing pervasive computing systems. Firstly, a formal modeling framework is proposed to cover the main characteristics of pervasive computing systems (e.g., context-awareness, concurrent communications, layered architectures). Secondly, we identify the safety requirements (e.g., free of deadlocks and conflicts etc.) and propose their specifications as safety and liveness properties. Finally, we demonstrate our ideas using a case study of a …
Information-Theoretic Multi-View Domain Adaptation,
2012
Singapore Management University
Information-Theoretic Multi-View Domain Adaptation, Pei Yang, Wei Gao, Qi Tan, Kam-Fai Wong
Research Collection School Of Computing and Information Systems
We use multiple views for cross-domain document classification. The main idea is to strengthen the views’ consistency for target data with source training data by identifying the correlations of domain-specific features from different domains. We present an Information-theoretic Multi-view Adaptation Model (IMAM) based on a multi-way clustering scheme, where word and link clusters can draw together seemingly unrelated domain-specific features from both sides and iteratively boost the consistency between document clusterings based on word and link views. Experiments show that IMAM significantly outperforms state-of-the-art baselines.
