Engineering Commons^™

Non-Trivial Off-Path Network Measurements Without Shared Side-Channel Resource Exhaustion, Geoffrey I. Alexander

Computer Science ETDs

Most traditional network measurement scans and attacks are carried out through the use of direct, on-path network packet transmission. This requires that a machine be on-path (i.e, involved in the packet transmission process) and as a result have direct access to the data packets being transmitted. This limits network scans and attacks to situations where access can be gained to an on-path machine. If, for example, a researcher wanted to measure the round trip time between two machines they did not have access to, traditional scans would be of little help as they require access to an on-path machine to …

Robot Motion Planning In Dynamic Environments, Hao-Tien Lewis Chiang

Computer Science ETDs

Robot motion planning in dynamic environments is critical for many robotic applications, such as self-driving cars, UAVs and service robots operating in changing environments. However, motion planning in dynamic environments is very challenging as this problem has been shown to be NP-Hard and in PSPACE, even in the simplest case. As a result, the lack of safe, efficient planning solutions for real-world robots is one of the biggest obstacles for ubiquitous adoption of robots in everyday life. Specifically, there are four main challenges facing motion planning in dynamic environments: obstacle motion uncertainty, obstacle interaction, complex robot dynamics and noise, and …

Using Uncertainty To Interpret Supervised Machine Learning Predictions, Michael C. Darling

Electrical and Computer Engineering ETDs

Traditionally, machine learning models are assessed using methods that estimate an average performance against samples drawn from a particular distribution. Examples include the use of cross-validation or hold0out to estimate classification error, F-score, precision, and recall.

While these measures provide valuable information, they do not tell us a model's certainty relative to particular regions of the input space. Typically there are regions where the model can differentiate the classes with certainty, and regions where the model is much less certain about its predictions.

In this dissertation we explore numerous approaches for quantifying uncertainty in the individual predictions made by supervised …

Artificial Intelligence Empowered Uavs Data Offloading In Mobile Edge Computing, Nicholas Alexander Kemp

Electrical and Computer Engineering ETDs

The advances introduced by Unmanned Aerial Vehicles (UAVs) are manifold and have paved the path for the full integration of UAVs, as intelligent objects, into the Internet of Things (IoT). This paper brings artificial intelligence into the UAVs data offloading process in a multi-server Mobile Edge Computing (MEC) environment, by adopting principles and concepts from game theory and reinforcement learning. Initially, the autonomous MEC server selection for partial data offloading is performed by the UAVs, based on the theory of the stochastic learning automata. A non-cooperative game among the UAVs is then formulated to determine the UAVs' data to be …

Thwarting Adversaries With Randomness And Irrationality, Abhinav Aggarwal

Computer Science ETDs

Distributed systems are ubiquitous today: from the Internet used by billions of people around the world to the small scale IoT devices. With this rapidly increasing need to perform computation at scales larger than ever before, comes the need to ensure resilience to adversarial failures so that these systems can continue to behave as intended even when some malicious tampering happens.

In this dissertation, we explore the power of randomness and the difficulty of rationally approximating the Golden Ratio to thwart adversarial behavior in two different problems in distributed computing: interactive communication and robust collaborative search. While randomness helps with …

An Efficient Multiple-Place Foraging Algorithm For Scalable Robot Swarms, Qi Lu

Computer Science ETDs

Searching and collecting multiple resources from large unmapped environments is an important challenge. It is particularly difficult given limited time, a large search area and incomplete data about the environment. This search task is an abstraction of many real-world applications such as search and rescue, hazardous material clean-up, and space exploration. The collective foraging behavior of robot swarms is an effective approach for this task. In our work, individual robots have limited sensing and communication range (like ants), but they are organized and work together to complete foraging tasks collectively. An efficient foraging algorithm coordinates robots to search and collect …

Large Scale Electronic Health Record Data And Echocardiography Video Analysis For Mortality Risk Prediction, Alvaro Emilio Ulloa Cerna

Electrical and Computer Engineering ETDs

Electronic health records contain the clinical history of patients. The enormous potential for discovery in such a rich dataset is hampered by their complexity. We hypothesize that machine learning models trained on EHR data can predict future clinical events significantly better than current models. We analyze an EHR database of 594,862 Echocardiography studies from 272,280 unique patients with both unsupervised and supervised machine learning techniques.

In the unsupervised approach, we first develop a simulation framework to evaluate a family of different clustering pipelines. We apply the optimized approach to 41,645 patients with heart failure without providing any survival information to …

Recipe For Disaster, Zac Travis

MFA Thesis Exhibit Catalogs

Today’s rapid advances in algorithmic processes are creating and generating predictions through common applications, including speech recognition, natural language (text) generation, search engine prediction, social media personalization, and product recommendations. These algorithmic processes rapidly sort through streams of computational calculations and personal digital footprints to predict, make decisions, translate, and attempt to mimic human cognitive function as closely as possible. This is known as machine learning.

The project Recipe for Disaster was developed by exploring automation in technology, specifically through the use of machine learning and recurrent neural networks. These algorithmic models feed on large amounts of data as a …