Filtering by
- All Subjects: Machine Learning
- Creators: Computer Science and Engineering Program
- Creators: Li, Jing
- Member of: Theses and Dissertations
System and software verification is a vital component in the development and reliability of cyber-physical systems - especially in critical domains where the margin of error is minimal. In the case of autonomous driving systems (ADS), the vision perception subsystem is a necessity to ensure correct maneuvering of the environment and identification of objects. The challenge posed in perception systems involves verifying the accuracy and rigidity of detections. The use of Spatio-Temporal Perception Logic (STPL) enables the user to express requirements for the perception system to verify, validate, and ensure its behavior; however, a drawback to STPL involves its accessibility. It is limited to individuals with an expert or higher-level knowledge of temporal and spatial logics, and the formal-written requirements become quite verbose with more restrictions imposed. In this thesis, I propose a domain-specific language (DSL) catered to Spatio-Temporal Perception Logic to enable non-expert users the ability to capture requirements for perception subsystems while reducing the necessity to have an experienced background in said logic. The domain-specific language for the Spatio-Temporal Perception Logic is built upon the formal language with two abstractions. The main abstraction captures simple programming statements that are translated to a lower-level STPL expression accepted by the testing monitor. The STPL DSL provides a seamless interface to writing formal expressions while maintaining the power and expressiveness of STPL. These translated equivalent expressions are capable of directing a standard for perception systems to ensure the safety and reduce the risks involved in ill-formed detections.
Robots are often used in long-duration scenarios, such as on the surface of Mars,where they may need to adapt to environmental changes. Typically, robots have been built specifically for single tasks, such as moving boxes in a warehouse or surveying construction sites. However, there is a modern trend away from human hand-engineering and toward robot learning. To this end, the ideal robot is not engineered,but automatically designed for a specific task. This thesis focuses on robots which learn path-planning algorithms for specific environments. Learning is accomplished via genetic programming. Path-planners are represented as Python code, which is optimized via Pareto evolution. These planners are encouraged to explore curiously and efficiently. This research asks the questions: “How can robots exhibit life-long learning where they adapt to changing environments in a robust way?”, and “How can robots learn to be curious?”.
In this paper, I introduce the fake news problem and detail how it has been exacerbated<br/>through social media. I explore current practices for fake news detection using natural language<br/>processing and current benchmarks in ranking the efficacy of various language models. Using a<br/>Twitter-specific benchmark, I attempt to reproduce the scores of six language models<br/>demonstrating their effectiveness in seven tweet classification tasks. I explain the successes and<br/>challenges in reproducing these results and provide analysis for the future implications of fake<br/>news research.
Human activity recognition is the task of identifying a person’s movement from sensors in a wearable device, such as a smartphone, smartwatch, or a medical-grade device. A great method for this task is machine learning, which is the study of algorithms that learn and improve on their own with the help of massive amounts of useful data. These classification models can accurately classify activities with the time-series data from accelerometers and gyroscopes. A significant way to improve the accuracy of these machine learning models is preprocessing the data, essentially augmenting data to make the identification of each activity, or class, easier for the model. <br/>On this topic, this paper explains the design of SigNorm, a new web application which lets users conveniently transform time-series data and view the effects of those transformations in a code-free, browser-based user interface. The second and final section explains my take on a human activity recognition problem, which involves comparing a preprocessed dataset to an un-augmented one, and comparing the differences in accuracy using a one-dimensional convolutional neural network to make classifications.
This paper is centered on the use of generative adversarial networks (GANs) to convert or generate RGB images from grayscale ones. The primary goal is to create sensible and colorful versions of a set of grayscale images by training a discriminator to recognize failed or generated images and training a generator to attempt to satisfy the discriminator. The network design is described in further detail below; however there are several potential issues that arise including the averaging of a color for certain images such that small details in an image are not assigned unique colors leading to a neutral blend. We attempt to mitigate this issue as much as possible.