Computer Engineering Commons^™

Feed Forward Neural Networks With Asymmetric Training, Archit Srivastava

Department of Computer Science and Engineering: Dissertations, Theses, and Student Research

Our work presents a new perspective on training feed-forward neural networks(FFNN). We introduce and formally define the notion of symmetry and asymmetry in the context of training of FFNN. We provide a mathematical definition to generalize the idea of sparsification and demonstrate how sparsification can induce asymmetric training in FFNN.

In FFNN, training consists of two phases, forward pass and backward pass. We define symmetric training in FFNN as follows-- If a neural network uses the same parameters for both forward pass and backward pass, then the training is said to be symmetric.

The definition of asymmetric training in artificial …

The Network Link Outlier Factor (Nlof) For Localizing Network Faults, Christopher Mendoza

Open Access Theses & Dissertations

This work presents the Network Link Outlier Factor (NLOF), a data analytics pipeline for network fault detection and localization solution that consists of four stages. In the first stage, flow record throughput values are clustered in two sub-stages: using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) and then a novel domain-specific ThroughPut Cluster (TPCluster) technique. In the second stage, Flow Outlier Factor (FOF) scores are computed for each flow. In the third stage, flows are traced onto the network. Finally, in the fourth stage, each link is given a Network Link Outlier Factor (NLOF) score which is the ratio …

Estimating Homophily In Social Networks Using Dyadic Predictions, George Berry, Antonio Sirianni, Ingmar Weber, Jisun An, Michael Macy

Research Collection School Of Computing and Information Systems

Predictions of node categories are commonly used to estimate homophily and other relational properties in networks. However, little is known about the validity of using predictions for this task. We show that estimating homophily in a network is a problem of predicting categories of dyads (edges) in the graph. Homophily estimates are unbiased when predictions of dyad categories are unbiased. Node-level prediction models, such as the use of names to classify ethnicity or gender, do not generally produce unbiased predictions of dyad categories and therefore produce biased homophily estimates. Bias comes from three sources: sampling bias, correlation between model errors …

Intelligent Networks For High Performance Computing, William Whitney Schonbein

Computer Science ETDs

There exists a resurgence of interest in `smart' network interfaces that can operate on data as it flows through a network. However, while smart capabilities have been expanding, what they can do for high-performance computing (HPC) is not well-understood. In this work, we advance our understanding of the capabilities and contributions of smart network interfaces to HPC. First, we show current offloaded message demultiplexing can mitigate (but not eliminate) overheads incurred by multithreaded communication. Second, we demonstrate current offloaded capabilities can be leveraged to provide Turing complete program execution on the interface. We elaborate with a framework for offloading arbitrary …

Improving Computer Network Operations Through Automated Interpretation Of State, Abhishek Dwaraki

Doctoral Dissertations

Networked systems today are hyper-scaled entities that provide core functionality for distributed services and applications spanning personal, business, and government use. It is critical to maintain correct operation of these networks to avoid adverse business outcomes. The advent of programmable networks has provided much needed fine-grained network control, enabling providers and operators alike to build some innovative networking architectures and solutions. At the same time, they have given rise to new challenges in network management. These architectures, coupled with a multitude of devices, protocols, virtual overlays on top of physical data-plane etc. make network management a highly challenging task. Existing …

Identification Of Users Via Ssh Timing Attack, Thomas J. Flucke

Master's Theses

Secure Shell, a tool to securely access and run programs on a remote machine, is an important tool for both system administrators and developers alike. The technology landscape is becoming increasingly distributed and reliant on tools such as Secure Shell to protect information as a user works on a system remotely. While Secure Shell accounts for the abuses the security of older tools such as telnet overlook, it still has fundamental vulnerabilities which leak information about both the user and their activities through timing attacks. The OpenSSH client, the implementation included in all Linux, Mac, and Windows computers, sends each …

Low-Precision Linear Algebra For Neural Networks, Frost Bennion Mitchell

Undergraduate Honors Capstone Projects

Neural networks have been widely responsible for recent advances in machine learning, powering technologies such as digital assistants and AR photography. LPLANN (Low-Precision Linear Algebra for Neural Networks) is a cross-platform library written in C++ used for implementing neural networks. The software allows users to set specific levels of precision for calculations. Low-precision calculations use advanced parallelization techniques (SIMD, SWAR) to run neural networks at faster rates than full-precision calculations. This library is lightweight enough to run on embedded systems, only relies on OpenMP as a dependency, and is portable to any operating system. LPLANN also includes optimizations to provide …

Optical Wireless Data Center Networks, Abdelbaset S. Hamza

Department of Computer Science and Engineering: Dissertations, Theses, and Student Research

Bandwidth and computation-intensive Big Data applications in disciplines like social media, bio- and nano-informatics, Internet-of-Things (IoT), and real-time analytics, are pushing existing access and core (backbone) networks as well as Data Center Networks (DCNs) to their limits. Next generation DCNs must support continuously increasing network traffic while satisfying minimum performance requirements of latency, reliability, flexibility and scalability. Therefore, a larger number of cables (i.e., copper-cables and fiber optics) may be required in conventional wired DCNs. In addition to limiting the possible topologies, large number of cables may result into design and development problems related to wire ducting and maintenance, heat …

Extension Of Cpe 454 Operating System, Joseph Nathaniel Arhar

Computer Science and Software Engineering

This project extended on the operating system I wrote in CPE 454 by adding additional features on top of the existing implementation. In order to implement them, I researched operating system design patterns and hardware details. I used wiki.osdev.org for most research, just like I did in CPE 454. The source code for the project is at https://github.com/josepharhar/jos.

Sd-Mcan: A Software-Defined Solution For Ip Mobility In Campus Area Networks, Adam Chase Calabrigo

Master's Theses

Campus Area Networks (CANs) are a subset of enterprise networks, comprised of a network core connecting multiple Local Area Networks (LANs) across a college campus. Traditionally, hosts connect to the CAN via a single point of attachment; however, the past decade has seen the employment of mobile computing rise dramatically. Mobile devices must obtain new Internet Protocol (IP) addresses at each LAN as they migrate, wasting address space and disrupting host services. To prevent these issues, modern CANs should support IP mobility: allowing devices to keep a single IP address as they migrate between LANs with low-latency handoffs. Traditional approaches …

Is Your Wireless Network Being Hacked?, Paul King, Ivan Smyth, Anthony Keane

The ITB Journal

Wireless networks provide vulnerable gateways for unauthorised entry to networks or even a standalone wireless computer. The independent radio signals that constitute wireless communications have no physical boundary to keep them in check. This allows a third party to easily eavesdrop on communications sessions and by capturing the data packets, they can break the encryption keys and access the data within the network. The public awareness of the insecurity of wireless networks is surprisingly poor despite frequent news media reports of the vulnerabilities of the equipment and the activities of the criminals prepare to exploit it. In this paper we …

Teaching The Internet Of Things: Bridging A Path From Cpe329, Steven Han, Rafael Lopez

Computer Engineering

“The ability to connect, communicate with, and remotely manage an incalculable number of networked, automated devices via the Internet is becoming pervasive, from the commercial kitchen to the residential basement room to the arm of the fitness buff.” - WSO2

In this report, we will investigate procedures and technologies used in IoT. A variety of cloud platforms will be described to demonstrate its strengths and usage on IoT applications. Furthermore, demonstrate the most popular hardware being used in several of these applications. This report is aimed to give a good understanding on what it takes to put together an IoT …

Making Sense Of Email Addresses On Drives, Neil C. Rowe, Riqui Schwamm, Michael R. Mccarrin, Ralucca Gera

Journal of Digital Forensics, Security and Law

Drives found during investigations often have useful information in the form of email addresses which can be acquired by search in the raw drive data independent of the file system. Using this data we can build a picture of the social networks that a drive owner participated in, even perhaps better than investigating their online profiles maintained by social-networking services because drives contain much data that users have not approved for public display. However, many addresses found on drives are not forensically interesting, such as sales and support links. We developed a program to filter these out using a Naïve …

Reducing Multiple Access Interference In Broadband Multi-User Wireless Networks, Ali Nayef Alqatawneh

Graduate Theses and Dissertations

This dissertation is devoted to developing multiple access interference (MAI) reduction techniques for multi-carrier multi-user wireless communication networks.

In multi-carrier code division multiple access (MC-CDMA) systems, a full multipath diversity can be achieved by transmitting one symbol over multiple orthogonal subcarriers by means of spreading codes. However, in frequency selective fading channels, orthogonality among users can be destroyed leading to MAI. MAI represents the main obstacle to support large number of users in multi-user wireless systems. Consequently, MAI reduction becomes a main challenge when designing multi-carrier multi-user wireless networks. In this dissertation, first, we study MC-CDMA systems with different existing …

Robustness And Prediction Accuracy Of Machine Learning For Objective Visual Quality Assessment, Andrew Hines, Paul Kendrick, Adriaan Barri, Manish Narwaria, Judith A. Redi

Conference papers

Machine Learning (ML) is a powerful tool to support the development of objective visual quality assessment metrics, serving as a substitute model for the perceptual mechanisms acting in visual quality appreciation. Nevertheless, the reliability of ML-based techniques within objective quality assessment metrics is often questioned. In this study, the robustness of ML in supporting objective quality assessment is investigated, specifically when the feature set adopted for prediction is suboptimal. A Principal Component Regression based algorithm and a Feed Forward Neural Network are compared when pooling the Structural Similarity Index (SSIM) features perturbed with noise. The neural network adapts better with …

On A Joint Physical Layer And Medium Access Control Sublayer Design For Efficient Wireless Sensor Networks And Applications, Mahir Lumumba Meghji

Theses: Doctorates and Masters

Wireless sensor networks (WSNs) are distributed networks comprising small sensing devices equipped with a processor, memory, power source, and often with the capability for short range wireless communication. These networks are used in various applications, and have created interest in WSN research and commercial uses, including industrial, scientific, household, military, medical and environmental domains. These initiatives have also been stimulated by the finalisation of the IEEE 802.15.4 standard, which defines the medium access control (MAC) and physical layer (PHY) for low-rate wireless personal area networks (LR-WPAN).

Future applications may require large WSNs consisting of huge numbers of inexpensive wireless sensor …

Openness, Intellectual Property And Standardization In The European Ict Sector, Carl Mair

IP Theory

No abstract provided.

Decision Strategies For A P2p Computing System, Grzegorz Chmaj, Krzysztof Walkowiak

Electrical & Computer Engineering Faculty Research

Peer-to-Peer (P2P) computing (also called ‘public-resource computing’) is an effective approach to perform computation of large tasks. Currently used P2P computing systems (e.g., BOINC) are most often centrally managed, i.e., the final result of computations is created at a central node using partial results – what may be not efficient in the case when numerous participants are willing to download the final result. In this paper, we propose a novel approach to P2P computing systems. We assume that results can be delivered to all peers in a distributed way using three types of network flows: unicast, Peer-to-Peer and anycast. We …

A Cross-Layer Modification To The Dsr Routing Protocol In Wireless Mesh Networks, Mustafa Ramadhan

Doctoral

A cross-layer modification to the DSR routing protocol that finds high throughput paths in WMNs has been introduced in this work. The Access Efficiency Factor (AEF) has been introduced in this modification as a local congestion avoidance metric for the DSR routing mechanism as an alternative to the hop count (Hc) metric. In this modification, the selected path is identified by finding a path with the highest minimum AEF (max_min_AEF) value. The basis of this study is to compare the performance of the Hc and max_min_AEF as routing metrics for the DSR protocol in WMNs using the OPNET modeler. Performance …

Dual-Use Of Compact Hf Radars For The Detection Of Mid- And Large-Size Vessels, Hugh Roarty, Don Barrick, Josh Kohut, Scott Glenn

Turkish Journal of Electrical Engineering and Computer Sciences

This paper describes the development of the SeaSonde High Frequency Radar system into a dual-use application for the mapping of ocean surface currents and detection of ships at sea. This development entailed the creation of a new radar waveform that would permit this dual-use as well as a detection algorithm to identify the ships in the radar spectra. The detection algorithm utilizes two methods for calculating a background signal level: an infinite impulse response (IIR) filter and a two-dimensional median filter. These two methods are employed simultaneously with multiple length averaging times to maximize the number of detections. The initial …

A Neural Network Approach To Border Gateway Protocol Peer Failure Detection And Prediction, Cory B. White

Master's Theses

The size and speed of computer networks continue to expand at a rapid pace, as do the corresponding errors, failures, and faults inherent within such extensive networks. This thesis introduces a novel approach to interface Border Gateway Protocol (BGP) computer networks with neural networks to learn the precursor connectivity patterns that emerge prior to a node failure. Details of the design and construction of a framework that utilizes neural networks to learn and monitor BGP connection states as a means of detecting and predicting BGP peer node failure are presented. Moreover, this framework is used to monitor a BGP network …

Data Security Measures In The It Service Industry: A Balance Between Knowledge & Action, N. Mlitwa, Y. Kachala

Journal of Digital Forensics, Security and Law

That “knowledge is power” is fast becoming a cliché within the intelligentsia. Such power however, depends largely on how knowledge itself is exchanged and used, which says a lot about the tools of its transmission, exchange, and storage. Information and communication technology (ICT) plays a significant role in this respect. As a networked tool, it enables efficient exchanges of video, audio and text data beyond geographical and time constraints. Since this data is exchanged over the worldwide web (www), it can be accessible by anyone in the world using the internet. The risk of unauthorised access, interception, modification, or even …

Reputed Authenticated Routing For Ad Hoc Networks Protocol (Reputed-Aran), Abdalla Ahmed Fekry Mahmoud

Archived Theses and Dissertations

A mobile ad hoc network (MANET) is a spontaneous network that can be established with no fixed infrastructure. This means that all its nodes behave as routers and take part in its discovery and maintenance of routes to other nodes in the network. Its routing protocol has to be able to cope with the new challenges that a MANET creates such as nodes mobility, security maintenance, quality of service, limited bandwidth and limited power supply. These challenges set new demands on MANET routing protocols. With the increasing interest in MANETs, there has been a greater focus on the subject of …

Characterization Of Critical Network Components Of Coupled Oscillators, Gregory Holifield

Electronic Theses and Dissertations

This dissertation analyzes the fundamental limits for the determination of the network structure of loosely coupled oscillators based on observing the behavior of the network, specifically, node synchronization. The determination of the requisite characteristics and underlying behaviors necessary for the application of a theoretical mechanism for determining the underlying network topology in a network of loosely coupled natural oscillators are the desired outcome. To that end, this effort defines an analytical framework where key components of networks of coupled oscillators are isolated in order to determine the relationships between the various components. The relationship between the number of nodes in …

Optimization Of Receptive Fields For Mlp Networks With Ensemble Encoding, Deena Osama Hassan

Archived Theses and Dissertations

Ensemble encoding is a biologically-motivated, distributed data representation scheme for MLP networks. Multiple overlapping receptive fields are used to enhance locality of representation. The number, form, and placement of receptive fields have a great impact on performance. This thesis presents four heuristics, two based on descriptive statistics, and two based on clustering, for optimizing receptive field configuration, and compares their performance on four benchmark data sets. The two statistical approaches are based on the mean and median properties of the data set. The two clustering methods are the c-means and fuzzy c-means clustering. The four data sets used are well-known …

Power-Aware On Demand Routing Protocol For Mobile Wireless Ad-Hoc Networks, Mohammed Kamel Fattouh Mohammed

Archived Theses and Dissertations

No abstract provided.

A Formal Object Model For Layered Networks To Support Verification And Simulation, Rasha M. B. E. Morsi

Electrical & Computer Engineering Theses & Dissertations

This work presents an abstract formal model of the interconnection structure of the Open Systems Interconnection Reference Model (OSI-RM) developed using Object-Oriented modeling principles permitting it to serve as a re-usable platform in supporting the development of simulations and formal methods applied to layered network protocols. A simulation of the object model using MODSIM III was developed and Prototype Verification System (PVS) was used to show the applicability of the object model to formal methods by formally specifying and verifying a Global Systems for Mobile communications (GSM) protocol. This application has proved to be successful in two aspects. The first …

Multicast Services For Multimedia Collaborative Applications, Emad Eldin Mohamed Mohamed

Computer Science Theses & Dissertations

This work aims at providing multicast services for multimedia collaborative applications over large inter-networks such as the Internet. Multimedia collaborative applications are typically of small group size, slow group membership dynamics, and awareness of participants' identities and locations. Moreover, they usually consist of several components such as audio, video, shared whiteboard, and single user application sharing engines that collectively help make the collaboration session successful. Each of these components has its demands from the communication layer that may differ from one component to another. This dissertation identifies the overall characteristics of multimedia collaborative applications and their individual components. It also …