Engineering Commons^™

Development Of A Design Guideline For Pile Foundations Subjected To Liquefaction-Induced Lateral Spreading, Milad Souri, Arash Khosravifar

Student Research Symposium

Past earthquakes confirmed that seismically induced kinematic loads from soil lateral spreading and inertial loads from structure can cause severe damages to pile foundations. The research questions are:

• How to combine inertial and kinematic loads in design of pile foundations in liquefied soil?
• How the combination of inertia and kinematics changes with depth?
• How this combination is affected by long-duration earthquakes?
• How this combination affects inelastic demands in piles?

An Assessment Of The Decision Making Units’ Efficiency In Service Systems (The Case Of Cellular Telecom), Maoloud Dabab, Timothy R. Anderson

Student Research Symposium

Most tools and models on performance and quality of service management are generic and do not solve the complex technical systems, which the most critical component on the network and where these tools should be applied. The objective of this research is to assess the cellular performance and Base Transceiver Station (BTS) efficiency by proposing a robust model that is derived from multiple Key Performance Indicators (KPIs) based on technical and financial aspects. The novelty of this research provides a comprehensive multidimensional model for tuning the BTS parameters, which can lead to developing a standard global mobile network KPI. The …

Explanation Methods For Neural Networks, Jack H. Chen, Christof Teuscher

Student Research Symposium

Neural Networks (NNs) have become a basis of almost all state-of-the-art machine learning algorithms and classifiers. While NNs have been shown to generalize well to real-world examples, researchers have struggled to show why they work on an intuitive level. We designed several methods to explain the decisions of two state-of-the-art NN classifiers, ResNet and an All-CNN, in the context of the Japanese Society of Radiological Technology (JSRT) lung nodule dataset and the CIFAR-10 image dataset. Leading explanation methods LIME and Grad-CAM generate variations of heat maps which represent the regions of the input determined salient by the NN. We analyze …

Exploring And Expanding The One-Pixel Attack, Umairullah Khan, Walt Woods, Christof Teuscher

Student Research Symposium

In machine learning research, adversarial examples are normal inputs to a classifier that have been specifically perturbed to cause the model to misclassify the input. These perturbations rarely affect the human readability of an input, even though the model’s output is drastically different. Recent work has demonstrated that image-classifying deep neural networks (DNNs) can be reliably fooled with the modification of a single pixel in the input image, without knowledge of a DNN’s internal parameters. This “one-pixel attack” utilizes an iterative evolutionary optimizer known as differential evolution (DE) to find the most effective pixel to perturb, via the evaluation of …

Diagnostic Imaging Of Structural Concrete Using Ground Penetrating Radar And Ultrasonic Array, Sina Mehdinia, Thomas Schumacher, Eric Wan, Xubo Song

Student Research Symposium

Structural concrete is the most widely used construction material in the world. Many structures critical to a society such as bridges, hospitals, and airports are built with concrete. While this material is well understood from a mechanical design point of view, still no accurate quantitative tools exist to assess it for damage and deterioration. This is of particular concern for an urban area like Portland with a mega-thrust earthquake waiting to occur. Non-destructive evaluation tools that can quickly and accurately give a full picture of the integrity of structural concrete elements will be key to help plan effective and safe …

Reliable Explanations Via Adversarial Examples On Robust Networks, Walt Woods, Jack H. Chen, Christof Teuscher

Student Research Symposium

Neural Networks (NNs) are increasingly used as the basis of advanced machine learning techniques in sensitive fields such as autonomous vehicles and medical imaging. However, NNs have been found vulnerable to a class of imperceptible attacks, called adversarial examples, which arbitrarily alter the output of the network. To close the schism between needing reliability in real-world applications and the fragility of NNs, we propose a new method for stabilizing networks, and show that as an added bonus, our technique results in reliable, high-fidelity explanations for the NN's decision. Compared to the state-of-the-art, this technique increased the area under the curve …