Numerical Analysis and Scientific Computing Commons^™

A Model Combining Self-Attention And Weight Sharing For Human Activity Recognition, Lun Ma, Yue Yang, Daihe Wang, Guisheng Liao, Xing Li

Journal of System Simulation

Abstract: With the prevalence of wearable devices, human activity recognition based on wearable sensor data has garnered significant attention. The central issue in this field is how to extract effective behavioral information from raw sensor data to form corresponding feature vectors. Currently, convolutional neural networks and recurrent neural networks have been widely utilized for feature extraction from multisensory data. However, these networks struggle to globally capture the crucial temporal features inherent of human activity over time. To address this, a multi-CNN-BiLSTM-self attention (Multi-CBSA) model based on self-attention and weight sharing has been proposed, taking into consideration the logical correlations among …

Adavis: Adaptive And Explainable Visualization Recommendation For Tabular Data, Songheng Zhang, Haotian Li, Huamin Qu, Yong Wang

Research Collection School Of Computing and Information Systems

Automated visualization recommendation facilitates the rapid creation of effective visualizations, which is especially beneficial for users with limited time and limited knowledge of data visualization. There is an increasing trend in leveraging machine learning (ML) techniques to achieve an end-to-end visualization recommendation. However, existing ML-based approaches implicitly assume that there is only one appropriate visualization for a specific dataset, which is often not true for real applications. Also, they often work like a black box, and are difficult for users to understand the reasons for recommending specific visualizations. To fill the research gap, we propose AdaVis, an adaptive and explainable …

Monocular Bev Perception Of Road Scenes Via Front-To-Top View Projection, Wenxi Liu, Qi Li, Weixiang Yang, Jiaxin Cai, Yuanhong Yu, Yuexin Ma, Shengfeng He, Jia Pan

Research Collection School Of Computing and Information Systems

HD map reconstruction is crucial for autonomous driving. LiDAR-based methods are limited due to expensive sensors and time-consuming computation. Camera-based methods usually need to perform road segmentation and view transformation separately, which often causes distortion and missing content. To push the limits of the technology, we present a novel framework that reconstructs a local map formed by road layout and vehicle occupancy in the bird's-eye view given a front-view monocular image only. We propose a front-to-top view projection (FTVP) module, which takes the constraint of cycle consistency between views into account and makes full use of their correlation to strengthen …

3d Streamline Visualization Method Based On Clustering Fusion, Xuqiang Shao, Ya Cheng, Yizhong Jin

Journal of System Simulation

Abstract: In order to solve the problems of incomplete feature extraction, continuity destruction of flow field by visual results, and poor representation of streamline caused by unstable clustering division when the clustering method is used to realize 3D streamline visualization. A 3D streamline visualization method based on clustering fusion is proposed. It consists of a distance measurement method between features and a clustering fusion method, which takes the inter-feature distance and spatial distance as the similarity between streamlines for clustering and then performs weighted merging and subdivision of the obtained clustering result. The method has been tested on data sets …

Efficient Unsupervised Video Hashing With Contextual Modeling And Structural Controlling, Jingru Duan, Yanbin Hao, Bin Zhu, Lechao Cheng, Pengyuan Zhou, Xiang Wang

Research Collection School Of Computing and Information Systems

The most important effect of the video hashing technique is to support fast retrieval, which is benefiting from the high efficiency of binary calculation. Current video hash approaches are thus mainly targeted at learning compact binary codes to represent video content accurately. However, they may overlook the generation efficiency for hash codes, i.e., designing lightweight neural networks. This paper proposes an method, which is not only for computing compact hash codes but also for designing a lightweight deep model. Specifically, we present an MLP-based model, where the video tensor is split into several groups and multiple axial contexts are explored …

Improving Rumor Detection By Promoting Information Campaigns With Transformer-Based Generative Adversarial Learning, Jing Ma, Jun Li, Wei Gao, Yang Yang, Kam-Fai Wong

Research Collection School Of Computing and Information Systems

Rumors can cause devastating consequences to individuals and our society. Analysis shows that the widespread of rumors typically results from deliberate promotion of information aiming to shape the collective public opinions on the concerned event. In this paper, we combat such chaotic phenomenon with a countermeasure by mirroring against how such chaos is created to make rumor detection more robust and effective. Our idea is inspired by adversarial learning method originated from Generative Adversarial Networks (GAN). We propose a GAN-style approach, where a generator is designed to produce uncertain or conflicting voices, further polarizing the original conversational threads to boost …

Dashboard Design Mining And Recommendation, Yanna Lin, Haotian Li, Aoyu Wu, Yong Wang, Huamin Qu

Research Collection School Of Computing and Information Systems

Dashboards, which comprise multiple views on a single display, help analyze and communicate multiple perspectives of data simultaneously. However, creating effective and elegant dashboards is challenging since it requires careful and logical arrangement and coordination of multiple visualizations. To solve the problem, we propose a data-driven approach for mining design rules from dashboards and automating dashboard organization. Specifically, we focus on two prominent aspects of the organization: , which describes the position, size, and layout of each view in the display space; and, which indicates the interaction between pairwise views. We build a new dataset containing 854 dashboards crawled online, …

Engine Wear Fault Diagnosis Based On Supervised Kernel Entropy Component Analysis, Zhichao Zhu, Dinghui Wu, Yuanchang Yue

Journal of System Simulation

Abstract: Focus on the influence of environment on engine operation, which leads to a large amount of redundant information and nonlinear structure in oil spectral data that affects the engine fault diagnosis results, the feature extraction method of SKECA (supervised kernel entropy component analysis) is proposed. A supervised learning algorithm is adopted on the basis of Kernel Entropy Component Analysis, which extracts the inherent geometric features of oil spectrum data to make the extracted fault features include the discriminative information. GA (genetic algorithm) is used to find parameters to optimize the results of feature extraction, and SVM (support vector machine) …

Action-Centric Relation Transformer Network For Video Question Answering, Jipeng Zhang, Jie Shao, Rui Cao, Lianli Gao, Xing Xu, Heng Tao Shen

Research Collection School Of Computing and Information Systems

Video question answering (VideoQA) has emerged as a popular research topic in recent years. Enormous efforts have been devoted to developing more effective fusion strategies and better intra-modal feature preparation. To explore these issues further, we identify two key problems. (1) Current works take almost no account of introducing action of interest in video representation. Additionally, there exists insufficient labeling data on where the action of interest is in many datasets. However, questions in VideoQA are usually action-centric. (2) Frame-to-frame relations, which can provide useful temporal attributes (e.g., state transition, action counting), lack relevant research. Based on these observations, we …

A Large Scale Study Of Long-Time Contributor Prediction For Github Projects, Lingfeng Bao, Xin Xia, David Lo, Gail C. Murphy

Research Collection School Of Computing and Information Systems

The continuous contributions made by long time contributors (LTCs) are a key factor enabling open source software (OSS) projects to be successful and survival. We study Github as it has a large number of OSS projects and millions of contributors, which enables the study of the transition from newcomers to LTCs. In this paper, we investigate whether we can effectively predict newcomers in OSS projects to be LTCs based on their activity data that is collected from Github. We collect Github data from GHTorrent, a mirror of Github data. We select the most popular 917 projects, which contain 75,046 contributors. …

A Data-Driven Method For Online Monitoring Tube Wall Thinning Process In Dynamic Noisy Environment, Chen Zhang, Jun Long Lim, Ouyang Liu, Aayush Madan, Yongwei Zhu, Shili Xiang, Kai Wu, Rebecca Yen-Ni Wong, Jiliang Eugene Phua, Karan M. Sabnani, Keng Boon Siah, Wenyu Jiang, Yixin Wang, Emily Jianzhong Hao, Hoi, Steven C. H.

Research Collection School Of Computing and Information Systems

Tube internal erosion, which corresponds to its wall thinning process, is one of the major safety concerns for tubes. Many sensing technologies have been developed to detect a tube wall thinning process. Among them, fiber Bragg grating (FBG) sensors are the most popular ones due to their precise measurement properties. Most of the current works focus on how to design different types of FBG sensors according to certain physical laws and only test their sensors in controlled laboratory conditions. However, in practice, an industrial system usually suffers from harsh and dynamic environmental conditions, and FBG signals are affected by many …

Find Me If You Can: Aligning Users In Different Social Networks, Priyanka Kasbekar, Katerina Potika, Chris Pollett

Faculty Publications, Computer Science

Online Social Networks allow users to share experiences with friends and relatives, make announcements, find news and jobs, and more. Several have user bases that number in the hundred of millions and even billions. Very often many users belong to multiple social networks at the same time under possibly different user names. Identifying a user from one social network on another social network gives information about a user's behavior on each platform, which in turn can help companies perform graph mining tasks, such as community detection and link prediction. The process of identifying or aligning users in multiple networks is …

Study On Hand Gesture Recognition And Portfolio Optimization Model Based On Svm, Zhiwei Cai, Shuyan Wu, Junfeng Song

Journal of System Simulation

Abstract: Hand gesture recognition was researched. The idea of extracting related features was proposed by using SVM algorithm in machine learning domain, and combination optimization method was used, which consists of ANN, HMM and DTW, to do hand gesture recognition. The experimental results show that portfolio optimization model based gesture recognition method has high accuracy and is very effective.

Heat Kernel Signature Extraction Algorithm Based On Mesh Simplification, Haisheng Li, Sun Li, Cai Qiang, Cao Jian

Journal of System Simulation

Abstract: Heat kernel signature has been proposed for 3D model feature extraction in recent years. However, the performance of heat kernel signature is inefficient, especially when the models have large number of vertices. Mesh simplification algorithm based on quadric error metrics was used to preprocess 3D model and the heat kernel signature was calculated based on the simplified model. Experiments show that the feature extracting time of the simplified model is less than the original model. The more vertices of the original model, the more obvious of the improved efficiency. The heat kernel signature of simplified model is consistent with …

Improved Method Of Extracting Hks Descriptors And Non-Rigid Classification Applications, Jingyu Jiang, Lili Wan

Journal of System Simulation

Abstract: In order to make the HKS（heat kernel signature）have wider applicability in non-rigid shape analysis, an improved method of extracting HKS descriptors for unconnected non-rigid 3D models was proposed. The largest connected component was obtained. The HKS descriptors of the largest connected component were calculated and those descriptors of the boundary vertices and their 1-ring neighbors were excluded. For shape classifications, the dictionary was learned for each class based on the sparse representation theory. For a test model, each dictionary was utilized to sparsely represent its descriptor set, and the most appropriate dictionary was determined by the representation error, …

Research Of Merging Three-Dimensional Static Scene And Moving Objects Based On Video, Kunjin He, Wang Lin, Jianxin Liu, Zhengming Chen, Xiaozhong Chen

Journal of System Simulation

Abstract: Against the difficulty in three-dimensional real-time rendering of scene that contains moving objects, a video-based method in building and merging three-dimensional static scene and moving objects was proposed. According to the classification of moving objects, a three-dimensional static scene was established and a parameterized model base was produced. Based on the video, moving objects were detected and features (including basic features and dynamic features) were extracted. Based on basic features, corresponding type model in the parameterized model base was instantiated. Based on the dynamic features, the three-dimensional static scene and moving object models were merged and the …

Research On Simulation Of Multi-Target Micro-Doppler Separation And Extraction In Ballistic Midcourse, Yizhe Wang, Cunqian Feng, Jingqing Li

Journal of System Simulation

Abstract: Aiming at the intricate overlap and difficult separation and extraction of micro-Doppler information in Doppler spectra of warheads and fragments in midcourse, a novel method based on CEEMD and improved self-adaptive Viterbi algorithm was proposed. By analyzing the differences of micro-Doppler distribution between warheads and fragments, the echo was decomposed by CEEMD and each IMF was denoised by wavelet threshold denoising method, resulting in separation of warheads and fragments echo. The fragments signal was stretched, the optimal path was extracted combined with improved self-adaptive Viterbi algorithm, and the separation of multi-target signal and extraction of micro-Doppler was realized. Simulation …

Motion Object Feature Extraction Method Based On Multi-Feature Fusion, Xidao Luan, Yuxiang Xie, Zhang Xin, Niu Xiao

Journal of System Simulation

Abstract: Motion object feature extraction is the basis of motion object classification. Traditionally motion object classification mainly depends on single feature extraction which is sensitive to the aspects like motion object detection area, angle, scale and noise disturbance, thus decreases the classification efficiency. To solve these problems and improve the robustness of the algorithms, a motion object feature extraction method based on multi-feature fusion was proposed. In this method, width height ratio feature, rotation invariant uniform local binary pattern feature and SIFT feature were considered, and by fusing them into the SVM and KNN classifier, motion object classification was carried …

Multi-Pose Pedestrian Detection Based On Posterior Multiple Sparse Dictionaries, Lingkang Gu, Mingzheng Zhou, Wang Jun, Xiu Yu

Journal of System Simulation

Abstract: In order to detect pedestrians effectively, a multi-pose pedestrian detection method based on posterior multiple sparse dictionaries was proposed. Through pre-learning multiple different sparse dictionaries, and sparse coding the image, statistics for each dictionary corresponds to sparse coding histogram as the pedestrian image feature descriptor. The common information of multiple sparse dictionary features of all positive samples was obtained, and the feature of a single pedestrian sample was weighted, and the features of a posteriori multiple sparse dictionary could be obtained. Then pedestrians of different poses and views were divided into subclasses with clustering algorithm. A classifier was trained …

Research On Augmented Reality Method Based On Unmarked Recognition, Li Qian, Shangbing Gao, Zhigeng Pan, Zhengwei Zhang, Chenghua Fang, Shengquan Wang

Journal of System Simulation

Abstract: As a kind of technology of superposing the virtual objects upon reality scene, AR (Augment Reality) expands the information quantity of the actual scene by virtual information. Aiming at the deficiency of previous AR application methods, the AR application method of unmarked recognition based on the optimized filtering was proposed in this paper. In this algorithm, filtering was used for image preprocessing to solve the poor matching efficiency problems of the past recognition algorithms, then three-dimensional model was established in two-dimensional images and the model was rendered in the two-dimensional plane. The experimental results show that, compared with …

Application Of Virtual Trial Makeup Based On Video, Jiayuan Liu, Jinfang Li, Hanwu He

Journal of System Simulation

Abstract: To try different cosmetic products in a convenient and low cost way, a virtual make-up algorithm based on a plane mesh model is proposed. The feature points are extracted from the face in the video by the Harr cascade classifier and the Dlib library. A mask pattern is built on the different parts of the face dynamically according to the texture coordinates of the plane mesh model. Theory is used to control the display area of the main texture of the planar mesh model. The mapping relationship between the main texture and the feature point coordinates of the planar …

Deep Air Learning: Interpolation, Prediction, And Feature Analysis Of Fine-Grained Air Quality, Zhongang Qi, Tianchun Wang, Guojie Song, Weisong Hu, Xi Li, Zhongfei Mark Zhang

Research Collection School Of Computing and Information Systems

The interpolation, prediction, and feature analysis of fine-gained air quality are three important topics in the area of urban air computing. The solutions to these topics can provide extremely useful information to support air pollution control, and consequently generate great societal and technical impacts. Most of the existing work solves the three problems separately by different models. In this paper, we propose a general and effective approach to solve the three problems in one model called the Deep Air Learning (DAL). The main idea of DAL lies in embedding feature selection and semi-supervised learning in different layers of the deep …

Who Will Leave The Company?: A Large-Scale Industry Study Of Developer Turnover By Mining Monthly Work Report, Lingfeng Bao, Zhenchang Xing, Xin Xia, David Lo, Shanping Li

Research Collection School Of Computing and Information Systems

Software developer turnover has become a big challenge for information technology (IT) companies. The departure of key software developers might cause big loss to an IT company since they also depart with important business knowledge and critical technical skills. Understanding developer turnover is very important for IT companies to retain talented developers and reduce the loss due to developers' departure. Previous studies mainly perform qualitative observations or simple statistical analysis of developers' activity data to understand developer turnover. In this paper, we investigate whether we can predict the turnover of software developers in non-open source companies by automatically analyzing monthly …

On Profiling Bots In Social Media, Richard J. Oentaryo, Arinto Murdopo, Philips K. Prasetyo, Ee Peng Lim

Research Collection School Of Computing and Information Systems

The popularity of social media platforms such as Twitter has led to the proliferation of automated bots, creating both opportunities and challenges in information dissemination, user engagements, and quality of services. Past works on profiling bots had been focused largely on malicious bots, with the assumption that these bots should be removed. In this work, however, we find many bots that are benign, and propose a new, broader categorization of bots based on their behaviors. This includes broadcast, consumption, and spam bots. To facilitate comprehensive analyses of bots and how they compare to human accounts, we develop a systematic profiling …

Computerized Classification Of Surface Spikes In Three-Dimensional Electron Microscopic Reconstructions Of Viruses, Younes Benkarroum

Dissertations, Theses, and Capstone Projects

The purpose of this research is to develop computer techniques for improved three-dimensional (3D) reconstruction of viruses from electron microscopic images of them and for the subsequent improved classification of the surface spikes in the resulting reconstruction. The broader impact of such work is the following.

Influenza is an infectious disease caused by rapidly-changing viruses that appear seasonally in the human population. New strains of influenza viruses appear every year, with the potential to cause a serious global pandemic. Two kinds of spikes – hemagglutinin (HA) and neuraminidase (NA) – decorate the surface of the virus particles and these proteins …

Where Is The Goldmine? Finding Promising Business Locations Through Facebook Data Analytics, Jovian Lin, Richard Oentaryo, Ee-Peng Lim, Casey Vu, Adrian Vu, Agus Kwee

Research Collection School Of Computing and Information Systems

If you were to open your own cafe, would you not want to effortlessly identify the most suitable location to set up your shop? Choosing an optimal physical location is a critical decision for numerous businesses, as many factors contribute to the final choice of the location. In this paper, we seek to address the issue by investigating the use of publicly available Facebook Pages data-which include user "check-ins", types of business, and business locations-to evaluate a user-selected physical location with respect to a type of business. Using a dataset of 20,877 food businesses in Singapore, we conduct analysis of …

Translation Initiation Sites Prediction With Mixture Gaussian Models In Human Cdna Sequences, G. Li, Tze-Yun Leong, Louxin Zhang

Research Collection School Of Computing and Information Systems

Translation initiation sites (TISs) are important signals in cDNA sequences. Many research efforts have tried to predict TISs in cDNA sequences. In this paper, we propose to use mixture Gaussian models for TIS prediction. Using both local features and some features generated from global measures, the proposed method predicts TISs with a sensitivity of 98 percent and a specificity of 93.6 percent. Our method outperforms many other existing methods in sensitivity while keeping specificity high. We attribute the improvement in sensitivity to the nature of the global features and the mixture Gaussian models. © 2005 IEEE.