Open Access. Powered by Scholars. Published by Universities.®
Physical Sciences and Mathematics Commons™
Open Access. Powered by Scholars. Published by Universities.®
- Keyword
-
- Color Images (2)
- Flexible AC transmission systems (2)
- Maximum Flow (2)
- Medical Image Processing (2)
- Power System (2)
-
- Power Transmission Control (2)
- Skin (2)
- 35 Mm Colour Photographic Slides (1)
- 5 Y (1)
- Adaptive Color Metric (1)
- Adaptive Replication (1)
- Approximation Algorithms (1)
- Artificial Neural Network Classification (1)
- Authorisation (1)
- Automated Cancer Detection (1)
- Batteries (1)
- Benign Tumors (1)
- Biological System Modeling (1)
- Cascading Failure (1)
- Computational Argumentation (1)
- Concurrent Process (1)
- Conflict Resolution (1)
- Coordinate Transformation (1)
- Deadliest Skin Cancer Form (1)
- Differential equations, Partial (1)
- Discriminant Features (1)
- Distributed Algorithm (1)
- Distributed Algorithms (1)
- Distributed Max-Flow (1)
- Electrical Flow Maximization (1)
Articles 1 - 13 of 13
Full-Text Articles in Physical Sciences and Mathematics
Cr-Sam: Curvature Regularized Sharpness-Aware Minimization, Tao Wu, Tony Tie Luo, Donald C. Wunsch
Cr-Sam: Curvature Regularized Sharpness-Aware Minimization, Tao Wu, Tony Tie Luo, Donald C. Wunsch
Computer Science Faculty Research & Creative Works
The Capacity to Generalize to Future Unseen Data Stands as One of the Utmost Crucial Attributes of Deep Neural Networks. Sharpness-Aware Minimization (SAM) Aims to Enhance the Generalizability by Minimizing Worst-Case Loss using One-Step Gradient Ascent as an Approximation. However, as Training Progresses, the Non-Linearity of the Loss Landscape Increases, Rendering One-Step Gradient Ascent Less Effective. on the Other Hand, Multi-Step Gradient Ascent Will Incur Higher Training Cost. in This Paper, We Introduce a Normalized Hessian Trace to Accurately Measure the Curvature of Loss Landscape on Both Training and Test Sets. in Particular, to Counter Excessive Non-Linearity of Loss Landscape, …
Lrs: Enhancing Adversarial Transferability Through Lipschitz Regularized Surrogate, Tao Wu, Tony Tie Luo, Donald C. Wunsch
Lrs: Enhancing Adversarial Transferability Through Lipschitz Regularized Surrogate, Tao Wu, Tony Tie Luo, Donald C. Wunsch
Computer Science Faculty Research & Creative Works
The Transferability of Adversarial Examples is of Central Importance to Transfer-Based Black-Box Adversarial Attacks. Previous Works for Generating Transferable Adversarial Examples Focus on Attacking Given Pretrained Surrogate Models While the Connections between Surrogate Models and Adversarial Trasferability Have Been overlooked. in This Paper, We Propose Lipschitz Regularized Surrogate (LRS) for Transfer-Based Black-Box Attacks, a Novel Approach that Transforms Surrogate Models towards Favorable Adversarial Transferability. using Such Transformed Surrogate Models, Any Existing Transfer-Based Black-Box Attack Can Run Without Any Change, Yet Achieving Much Better Performance. Specifically, We Impose Lipschitz Regularization on the Loss Landscape of Surrogate Models to Enable a Smoother …
Distributed Power Balancing For The Freedm System, Rav Akella, Fanjun Meng, Derek Ditch, Bruce M. Mcmillin, Mariesa Crow
Distributed Power Balancing For The Freedm System, Rav Akella, Fanjun Meng, Derek Ditch, Bruce M. Mcmillin, Mariesa Crow
Computer Science Faculty Research & Creative Works
The FREEDM microgrid is a test bed for a smart grid integrated with Distributed Grid Intelligence (DGI) to efficiently manage the distribution and storage of renewable energy. Within the FREEDM system, DGI applies distributed algorithms in a unique way to achieve economically feasible utilization and storage of alternative energy sources in a distributed fashion. The FREEDM microgrid consists of residential or industrial nodes with each node running a portion of the DGI process called Intelligent Energy Management (IEM). Such IEM nodes within FREEDM coordinate among themselves to efficiently and economically manage their power generation, utility and storage. Among a variety …
An Open Framework For Highly Concurrent Real-Time Hardware-In-The-Loop Simulation, Ryan C. Underwood, Bruce M. Mcmillin, Mariesa Crow
An Open Framework For Highly Concurrent Real-Time Hardware-In-The-Loop Simulation, Ryan C. Underwood, Bruce M. Mcmillin, Mariesa Crow
Computer Science Faculty Research & Creative Works
Hardware-in-the-loop (HIL) real-time simulation is becoming a significant tool in prototyping complex, highly available systems. The HIL approach permits testing of hardware prototypes of components that would be extremely costly or difficult to test in the deployed environment. In power system simulation, key issues are the ability to wrap the systems of equations (such as Partial Differential Equations) describing the deployed environment into real-time software models, provide low synchronization overhead between the hardware and software, and reduce reliance on proprietary platforms. This paper introduces an open source HIL simulation framework that can be ported to any standard Unix-like system on …
Use Of Max-Flow On Facts Devices, Adam Lininger, Bruce M. Mcmillin, Badrul H. Chowdhury, Mariesa Crow
Use Of Max-Flow On Facts Devices, Adam Lininger, Bruce M. Mcmillin, Badrul H. Chowdhury, Mariesa Crow
Computer Science Faculty Research & Creative Works
FACTS devices can be used to mitigate cascading failures in a power grid by controlling the power flow in individual lines. Placement and control are significant issues. We present a procedure for determining whether a scenario can be mitigated using the concept of maximum flow. If it can be mitigated, we determine what placement and control setting will solve the scenario. This paper treats fourteen cascading failure scenarios and reports on the use of the max-flow algorithm both in determining the mitigation of each scenario and in finding FACTS settings that will mitigate the scenario.
Management Of An Intelligent Argumentation Network For A Web-Based Collaborative Engineering Design Environment, Xiaoqing Frank Liu, Man Zheng, Ganesh K. Venayagamoorthy, Ming-Chuan Leu
Management Of An Intelligent Argumentation Network For A Web-Based Collaborative Engineering Design Environment, Xiaoqing Frank Liu, Man Zheng, Ganesh K. Venayagamoorthy, Ming-Chuan Leu
Computer Science Faculty Research & Creative Works
Conflict resolution is one of the most challenging tasks in collaborative engineering design. In our previous research, a web-based intelligent collaborative system was developed to address this challenge based on intelligent computational argumentation. However, two important issues were not resolved in that system: priority of participants and self-conflicting arguments. In this paper, we develop two methods for incorporating priorities of participants into the computational argumentation network: 1) weighted summation and 2) re-assessment of strengths of arguments based on priority of owners of the argument using fuzzy logic inference. In addition, we develop a method for detection of self-conflicting arguments. Incorporation …
An Instance-Based Structured Object Oriented Method For Co-Analysis/Co-Design Of Concurrent Embedded Systems, Matt Ryan, Xiaoqing Frank Liu, Bruce M. Mcmillin, Ying Cheng, Sule Simsek
An Instance-Based Structured Object Oriented Method For Co-Analysis/Co-Design Of Concurrent Embedded Systems, Matt Ryan, Xiaoqing Frank Liu, Bruce M. Mcmillin, Ying Cheng, Sule Simsek
Computer Science Faculty Research & Creative Works
The current object-oriented class-based approaches to hardware/software co-analysis/co-design of embedded systems are limited in their abilities to properly capture the structure of individual instances of hardware and software components and their interactions. This paper discusses a methodology to extend a structured objectoriented hardware/software co-design methodology based on the High Order Object-oriented Modeling Technique (HOOMT) to incorporate instance-based object and behavioral models. The instance-based structured object-oriented methodology will enable description of a system's structure based on individual instances of hardware and software components and specification of the interactions among them. In addition, lattices are introduced to specify the concurrent behavior of …
Structured Object-Oriented Co-Analysis/Co-Design Of Hardware/Software For The Facts Powers System, Matt Ryan, Sojan Markose, Xiaoqing Frank Liu, Bruce M. Mcmillin
Structured Object-Oriented Co-Analysis/Co-Design Of Hardware/Software For The Facts Powers System, Matt Ryan, Sojan Markose, Xiaoqing Frank Liu, Bruce M. Mcmillin
Computer Science Faculty Research & Creative Works
There are several approaches to the hardware/software design in embedded systems, ranging from the traditional sequential methods which focus on the determination of the hardware architecture prior to software design, to newer object-oriented approaches that attempt to apply software engineering methods to hardware design without a systematic process. This paper discusses a structured object-oriented methodology for the integrated co-analysis and co-design of hardware/software systems using an extended high order object-oriented modeling technique (HOOMT). This methodology offers a uniform method for hardware and software developers to jointly develop the specifications for and partitioning of the hardware and software components of a …
Power Transmission Control Using Distributed Max-Flow, Bruce M. Mcmillin, Austin Armbruster, Mariesa Crow, Michael R. Gosnell
Power Transmission Control Using Distributed Max-Flow, Bruce M. Mcmillin, Austin Armbruster, Mariesa Crow, Michael R. Gosnell
Computer Science Faculty Research & Creative Works
Existing maximum flow algorithms use one processor for all calculations or one processor per vertex in a graph to calculate the maximum possible flow through a graph's vertices. This is not suitable for practical implementation. We extend the max-flow work of Goldberg and Tarjan to a distributed algorithm to calculate maximum flow where the number of processors is less than the number of vertices in a graph. Our algorithm is applied to maximizing electrical flow within a power network where the power grid is modeled as a graph. Error detection measures are included to detect problems in a simulated power …
Adaptive Replication And Access Control Of Multimedia Data In A P2p Environment, Sanjay Kumar Madria, Sanjeev Agarwal
Adaptive Replication And Access Control Of Multimedia Data In A P2p Environment, Sanjay Kumar Madria, Sanjeev Agarwal
Computer Science Faculty Research & Creative Works
This paper explores some of the ideas and solutions related to replication and access control of multimedia data in a hierarchical P2P environment. We provided overview of the techniques to generate multiresolution of multimedia data and explored error recovery and access control issues.
A Distributed Discrete-Time Neural Network Architecture For Pattern Allocation And Control, A.T. Chronopoulos, Jagannathan Sarangapani
A Distributed Discrete-Time Neural Network Architecture For Pattern Allocation And Control, A.T. Chronopoulos, Jagannathan Sarangapani
Computer Science Faculty Research & Creative Works
No abstract provided.
Neural Network Diagnosis Of Malignant Melanoma From Color Images, Fikret Erçal, Hsi-Chieh Lee, William V. Stoecker, Randy Hays Moss, Anurag Chawla
Neural Network Diagnosis Of Malignant Melanoma From Color Images, Fikret Erçal, Hsi-Chieh Lee, William V. Stoecker, Randy Hays Moss, Anurag Chawla
Computer Science Faculty Research & Creative Works
Malignant melanoma is the deadliest form of all skin cancers. Approximately 32,000 new cases of malignant melanoma were diagnosed in 1991 in the United States, with approximately 80% of patients expected to survive 5 years. Fortunately, if detected early, even malignant melanoma may be treated successfully, Thus, in recent years, there has been rising interest in the automated detection and diagnosis of skin cancer, particularly malignant melanoma. Here, the authors present a novel neural network approach for the automated separation of melanoma from 3 benign categories of tumors which exhibit melanoma-like characteristics. The approach uses discriminant features, based on tumor …
Detection Of Skin Tumor Boundaries In Color Images, Fikret Erçal, M. Moganti, William V. Stoecker, Randy Hays Moss
Detection Of Skin Tumor Boundaries In Color Images, Fikret Erçal, M. Moganti, William V. Stoecker, Randy Hays Moss
Computer Science Faculty Research & Creative Works
A simple and yet effective method for finding the borders of tumors is presented as an initial step towards the diagnosis of skin tumors from their color images. The method makes use of an adaptive color metric from the red, green, and blue planes that contains information for discriminating the tumor from the background. Using this suitable coordinate transformation, the image is segmented. The tumor portion is then extracted from the segmented image and borders are drawn. Experimental results that verify the effectiveness of this approach are given