Theory and Algorithms Commons^™

Power-Efficient And Highly Scalable Parallel Graph Sampling Using Fpgas, Usman Tariq, Umer Cheema, Fahad Saeed

Parallel Computing and Data Science Lab Technical Reports

Energy efficiency is a crucial problem in data centers where big data is generally represented by directed or undirected graphs. Analysis of this big data graph is challenging due to volume and velocity of the data as well as irregular memory access patterns. Graph sampling is one of the most effective ways to reduce the size of graph while maintaining crucial characteristics. In this paper we present design and implementation of an FPGA based graph sampling method which is both time- and energy-efficient. This is in contrast to existing parallel approaches which include memory-distributed clusters, multicore and GPUs. Our …

High-Order Relaxed Multirate Infinitesimal Step Methods For Multiphysics Applications, Jean Sexton

Mathematics Theses and Dissertations

In this work, we consider numerical methods for integrating multirate ordinary differential equations. We are interested in the development of new multirate methods with good stability properties and improved efficiency over existing methods. We discuss the development of multirate methods, particularly focusing on those that are based on Runge-Kutta theory. We introduce the theory of Generalized Additive Runge-Kutta methods proposed by Sandu and Günther. We also introduce the theory of Recursive Flux Splitting Multirate Methods with Sub-cycling described by Schlegel, as well as the Multirate Infinitesimal Step methods this work is based on. We propose a generic structure called Flexible …

Sol: A Library For Scalable Online Learning Algorithms, Yue Wu, Steven C. H. Hoi, Chenghao Liu, Jing Lu, Doyen Sahoo, Nenghai Yu

Research Collection School Of Computing and Information Systems

SOL is an open-source library for scalable online learning with high-dimensional data. The library provides a family of regular and sparse online learning algorithms for large-scale classification tasks with high efficiency, scalability, portability, and extensibility. We provide easy-to-use command-line tools, python wrappers and library calls for users and developers, and comprehensive documents for both beginners and advanced users. SOL is not only a machine learning toolbox, but also a comprehensive experimental platform for online learning research. Experiments demonstrate that SOL is highly efficient and scalable for large-scale learning with high-dimensional data.

Benchmarking Single-Image Reflection Removal Algorithms, Renjie Wan, Boxin Shi, Ling-Yu Duan, Ah-Hwee Tan, Alex C. Kot

Research Collection School Of Computing and Information Systems

Removing undesired reflections from a photo taken in front of a glass is of great importance for enhancing the efficiency of visual computing systems. Various approaches have been proposed and shown to be visually plausible on small datasets collected by their authors. A quantitative comparison of existing approaches using the same dataset has never been conducted due to the lack of suitable benchmark data with ground truth. This paper presents the first captured Single-image Reflection Removal dataset `SIR 2 ' with 40 controlled and 100 wild scenes, ground truth of background and reflection. For each controlled scene, we further provide …

Methods For Real-Time Prediction Of The Mode Of Travel Using Smartphone-Based Gps And Accelerometer Data, Bryan D. Martin, Vittorio Addona, Julian Wolfson, Gediminas Adomavicius, Yingling Fan

Faculty Publications

We propose and compare combinations of several methods for classifying transportation activity data from smartphone GPS and accelerometer sensors. We have two main objectives. First, we aim to classify our data as accurately as possible. Second, we aim to reduce the dimensionality of the data as much as possible in order to reduce the computational burden of the classification. We combine dimension reduction and classification algorithms and compare them with a metric that balances accuracy and dimensionality. In doing so, we develop a classification algorithm that accurately classifies five different modes of transportation (i.e., walking, biking, car, bus and rail) …

Approximation Algorithms For Effective Team Formation, George Rabanca

Dissertations, Theses, and Capstone Projects

This dissertation investigates the problem of creating multiple disjoint teams of maximum efficacy from a fixed set of workers. We identify three parameters which directly correlate to the team effectiveness — team expertise, team cohesion and team size — and propose efficient algorithms for optimizing each in various settings. We show that under standard assumptions the problems we explore are not optimally solvable in polynomial time, and thus we focus on developing efficient algorithms with guaranteed worst case approximation bounds. First, we investigate maximizing team expertise in a setting where each worker has different expertise for each job and each …

A Combinatorial Framework For Multiple Rna Interaction Prediction, Syed Ali Ahmed

Dissertations, Theses, and Capstone Projects

The interaction of two RNA molecules involves a complex interplay between folding and binding that warranted recent developments in RNA-RNA interaction algorithms. However, biological mechanisms in which more than two RNAs take part in an interaction also exist.

A typical algorithmic approach to such problems is to find the minimum energy structure. Often the computationally optimal solution does not represent the biologically correct structure of the interaction. In addition, different biological structures may be observed, depending on several factors. Furthermore, scoring techniques often miss critical details about dependencies within different parts of the structure, which typically leads to lower scores …

Morphogenesis And Growth Driven By Selection Of Dynamical Properties, Yuri Cantor

Dissertations, Theses, and Capstone Projects

Organisms are understood to be complex adaptive systems that evolved to thrive in hostile environments. Though widely studied, the phenomena of organism development and growth, and their relationship to organism dynamics is not well understood. Indeed, the large number of components, their interconnectivity, and complex system interactions all obscure our ability to see, describe, and understand the functioning of biological organisms.

Here we take a synthetic and computational approach to the problem, abstracting the organism as a cellular automaton. Such systems are discrete digital models of real-world environments, making them more accessible and easier to study then their physical world …

Micro-Review Synthesis For Multi-Entity Summarization, Thanh-Son Nguyen, Hady W. Lauw, Panayiotis Tsaparas

Research Collection School Of Computing and Information Systems

Location-based social networks (LBSNs), exemplified by Foursquare, are fast gaining popularity. One important feature of LBSNs is micro-review. Upon check-in at a particular venue, a user may leave a short review (up to 200 characters long), also known as a tip. These tips are an important source of information for others to know more about various aspects of an entity (e.g., restaurant), such as food, waiting time, or service. However, a user is often interested not in one particular entity, but rather in several entities collectively, for instance within a neighborhood or a category. In this paper, we address the …

Ancr—An Adaptive Network Coding Routing Scheme For Wsns With Different-Success-Rate Links †, Xiang Ji, Anwen Wang, Chunyu Li, Chun Ma, Yao Peng, Dajin Wang, Qingyi Hua, Feng Chen, Dingyi Fang

Department of Computer Science Faculty Scholarship and Creative Works

As the underlying infrastructure of the Internet of Things (IoT), wireless sensor networks (WSNs) have been widely used in many applications. Network coding is a technique in WSNs to combine multiple channels of data in one transmission, wherever possible, to save node’s energy as well as increase the network throughput. So far most works on network coding are based on two assumptions to determine coding opportunities: (1) All the links in the network have the same transmission success rate; (2) Each link is bidirectional, and has the same transmission success rate on both ways. However, these assumptions may not be …

Parallelization Of Molecular Docking Algorithms Using Cuda For Use In Drug Discovery, Brandon Stewart, Jonathan Fine, Gaurav Chopra Phd

The Summer Undergraduate Research Fellowship (SURF) Symposium

Traditional drug discovery methodology uses a multitude of software packages to design and evaluate new drug-like compounds. While software packages implement a wide variety of methods, the serial (i.e. single core) implementation for many of these algorithms, prohibit large scale docking, such as proteome-wide docking (i.e. thousands of compounds with thousands of proteins). Several docking algorithms can be parallelized, significantly reducing the runtime of the calculations, thus enabling large-scale docking. Implementing algorithms that take advantage of the distributed nature of graphical processing units (GPUs) via the Compute Unified Device Architecture (CUDA) enables us to efficiently implement massively parallel algorithms. Two …

Vertex Weighted Spectral Clustering, Mohammad Masum

Electronic Theses and Dissertations

Spectral clustering is often used to partition a data set into a specified number of clusters. Both the unweighted and the vertex-weighted approaches use eigenvectors of the Laplacian matrix of a graph. Our focus is on using vertex-weighted methods to refine clustering of observations. An eigenvector corresponding with the second smallest eigenvalue of the Laplacian matrix of a graph is called a Fiedler vector. Coefficients of a Fiedler vector are used to partition vertices of a given graph into two clusters. A vertex of a graph is classified as unassociated if the Fiedler coefficient of the vertex is close to …

Investigating Genetic Algorithm Optimization Techniques In Video Games, Nathan Ambuehl

Undergraduate Honors Theses

Immersion is essential for player experience in video games. Artificial Intelligence serves as an agent that can generate human-like responses and intelligence to reinforce a player’s immersion into their environment. The most common strategy involved in video game AI is using decision trees to guide chosen actions. However, decision trees result in repetitive and robotic actions that reflect an unrealistic interaction. This experiment applies a genetic algorithm that explores selection, crossover, and mutation functions for genetic algorithm implementation in an isolated Super Mario Bros. pathfinding environment. An optimized pathfinding AI can be created by combining an elitist selection strategy with …

Large-Scale Online Feature Selection For Ultra-High Dimensional Sparse Data, Yue Wu, Steven C. H. Hoi, Tao Mei, Nenghai Yu

Research Collection School Of Computing and Information Systems

Feature selection (FS) is an important technique in machine learning and data mining, especially for large scale high-dimensional data. Most existing studies have been restricted to batch learning, which is often inefficient and poorly scalable when handling big data in real world. As real data may arrive sequentially and continuously, batch learning has to retrain the model for the new coming data, which is very computationally intensive. Online feature selection (OFS) is a promising new paradigm that is more efficient and scalable than batch learning algorithms. However, existing online algorithms usually fall short in their inferior efficacy. In this article, …

Sparse Online Learning Of Image Similarity, Xingyu Gao, Steven C. H. Hoi, Yongdong Zhang, Jianshe Zhou, Ji Wan, Zhenyu Chen, Jintao Li, Jianke Zhu

Research Collection School Of Computing and Information Systems

Learning image similarity plays a critical role in real-world multimedia information retrieval applications, especially in Content-Based Image Retrieval (CBIR) tasks, in which an accurate retrieval of visually similar objects largely relies on an effective image similarity function. Crafting a good similarity function is very challenging because visual contents of images are often represented as feature vectors in high-dimensional spaces, for example, via bag-of-words (BoW) representations, and traditional rigid similarity functions, for example, cosine similarity, are often suboptimal for CBIR tasks. In this article, we address this fundamental problem, that is, learning to optimize image similarity with sparse and high-dimensional representations …

Modeling Trajectories With Recurrent Neural Networks, Hao Wu, Ziyang Chen, Weiwei Sun, Baihua Zheng, Wei Wang

Research Collection School Of Computing and Information Systems

Modeling trajectory data is a building block for many smart-mobility initiatives. Existing approaches apply shallow models such as Markov chain and inverse reinforcement learning to model trajectories, which cannot capture the long-term dependencies. On the other hand, deep models such as Recurrent Neura lNetwork (RNN) have demonstrated their strength of modeling variable length sequences. However, directly adopting RNN to model trajectories is not appropriate because of the unique topological constraints faced by trajectories. Motivated by these findings, we design two RNN-based models which can make full advantage of the strength of RNN to capture variable length sequence and meanwhile to …

Accelerating Dynamic Graph Analytics On Gpus, Mo Shan, Yuchen Li, Bingsheng He, Kian-Lee Tan

Research Collection School Of Computing and Information Systems

As graph analytics often involves compute-intensive operations,GPUs have been extensively used to accelerate the processing. However, in many applications such as social networks, cyber security, and fraud detection, their representative graphs evolve frequently and one has to perform are build of the graph structure on GPUs to incorporate the updates. Hence, rebuilding the graphs becomes the bottleneck of processing high-speed graph streams. In this paper,we propose a GPU-based dynamic graph storage scheme to support existing graph algorithms easily. Furthermore,we propose parallel update algorithms to support efficient stream updates so that the maintained graph is immediately available for high-speed analytic processing …

Projection-Free Distributed Online Learning In Networks, Wenpeng Zhang, Peilin Zhao, Wenwu Zhu, Steven C. H. Hoi, Tong Zhang

Research Collection School Of Computing and Information Systems

The conditional gradient algorithm has regained a surge of research interest in recent years due to its high efficiency in handling large-scale machine learning problems. However, none of existing studies has explored it in the distributed online learning setting, where locally light computation is assumed. In this paper, we fill this gap by proposing the distributed online conditional gradient algorithm, which eschews the expensive projection operation needed in its counterpart algorithms by exploiting much simpler linear optimization steps. We give a regret bound for the proposed algorithm as a function of the network size and topology, which will be smaller …

Object Detection Meets Knowledge Graphs, Yuan Fang, Kingsley Kuan, Jie Lin, Cheston Tan, Vijay Chandrasekhar

Research Collection School Of Computing and Information Systems

Object detection in images is a crucial task in computer vision, with important applications ranging from security surveillance to autonomous vehicles. Existing state-of-the-art algorithms, including deep neural networks, only focus on utilizing features within an image itself, largely neglecting the vast amount of background knowledge about the real world. In this paper, we propose a novel framework of knowledge-aware object detection, which enables the integration of external knowledge such as knowledge graphs into any object detection algorithm. The framework employs the notion of semantic consistency to quantify and generalize knowledge, which improves object detection through a re-optimization process to achieve …

Online Multitask Relative Similarity Learning, Shuji Hao, Peilin Zhao, Yong Liu, Steven C. H. Hoi, Chunyan Miao

Research Collection School Of Computing and Information Systems

Relative similarity learning (RSL) aims to learn similarity functions from data with relative constraints. Most previous algorithms developed for RSL are batch-based learning approaches which suffer from poor scalability when dealing with real world data arriving sequentially. These methods are often designed to learn a single similarity function for a specific task. Therefore, they may be sub-optimal to solve multiple task learning problems. To overcome these limitations, we propose a scalable RSL framework named OMTRSL (Online Multi-Task Relative Similarity Learning). Specifically, we first develop a simple yet effective online learning algorithm for multi-task relative similarity learning. Then, we also propose …