This paper addresses the high model complexity and overconfident frame labeling of state-of-the-art (SOTA) action segmenters. Their complexity is typically justified by the need to sequentially refine action segmentation through multiple stages of a deep architecture. However, this multistage refinement does not take into account uncertainty of frame labeling predicted...
Distributed version control allows developers to manage software evolution among distributed development teams. But it does not eliminate all consistency and concurrency issues, and instead introduces additional complexity when merging code. And resolving merge conflicts is nontrivial when automated merging fails. In such cases, developers are forced to inspect the...
Metric spaces (X, d) are ubiquitous objects in mathematics and computer science that are able to capture pairwise distance relationships d(x, y) between points x, y ∈ X. Because of this, it is natural to ask what useful generalizations there are of metric spaces for capturing “k-wise distance relationships” d(x1,...
Explainable Artificial Intelligence (XAI) systems aim to improve users’ understanding of AI but rarely consider the inclusivity aspects of XAI. Without inclusive approaches, improving explanations might not work well for everyone. This study investigates leveraging users’ diverse problem-solving styles as an inclusive strategy to fix an XAI prototype, with the...
We present student perceptions of a new first-year engineering programming class that was designed by informed research practices. While the College of Engineering at Oregon State University saw a lot of major switching in the first year, there were not many students switching into computer science (CS). This could have...
Autonomous robotic agents are on their way to becoming in-home personal assistants, construction assistants, and warehouse workers. The degree of autonomy of such systems is reflected by the manner in which we specify goals to them; the abstraction of low-level commands to high-level goals goes hand-in-hand with increased autonomy. In...
Learning to recognize objects is a fundamental and essential step in human perception and understanding of the world. Accordingly, research of object discovery across diverse modalities plays a pivotal role in the context of computer vision. This field not only contributes significantly to enhancing our understanding of visual information but...
In this thesis, we propose a systematic code for correcting t = 1 insertion/deletion errors of the character ”0” that can occur between any two consecutive 1’s in a binary string. The code requires balanced input strings, where each word of length n contains ⌈n/2⌉ 0’s and ⌊n/2⌋ 1’s. This...
Emerging research shows that individual differences in how people use technology sometimes cluster by socioeconomic status (SES) and that when technology is not socioeconomically inclusive, low-SES individuals may abandon it. To understand how to improve technology’s SES-inclusivity, we present a multi-phase case study on SocioEconomicMag (SESMag), an emerging inspection method...
A secret sharing scheme allows a dealer to distribute a secret with a set of parties, such that only a certain subset of parties can collaborate and learn the shared secret. Traditional secret sharing schemes have been used as building blocks in various subdomains of cryptography. Recently, two new extensions...
One of the pervasive problems arising in our modern, digital world surrounds data breaches where an adversary, through zero-day exploitations, phishing, or old-fashioned social engineering attacks, gains access to a service’s data stores. Our society increasingly relies on these cloud-based services for everything from our taxes to personal communication. As...
While digital inclusivity researchers and software practitioners have been trying to address exclusion biases in Windows, Icons, Menus, and Pointers (WIMP) user interfaces (UIs) for a long time, little has been done to investigate if and how inclusive software design and its methods that have been devised for WIMP UIs...
Over time, Open Source Software (OSS) has become indispensable in the creation and upkeep of software products, serving as the fundamental building block for widely used solutions in our daily lives, including applications that enable communication, entertainment, and productivity. A sustainable OSS ecosystem is one that attracts and retains a...
In an increasingly computation-driven world, algorithms and mathematical models significantly impact decision making across various fields. To foster trust and understanding, it is crucial to provide users with clear and concise explanations of the reasoning behind the results produced by computational tools, especially when recommendations appear counterintuitive. Legal frameworks in...
In the ever-evolving field of computer science (CS) education, the significance of teachers and their backgrounds have often been overshadowed by the predominant focus on students. Teachers in the K-12 often lack the necessary expertise and have limited support provided by existing CS-based curricula. While research on CS education effectiveness...
The understanding of Discipline-Specific Language is an important competency for students of any field to begin mastering early in their studies, since it serves as a prerequisite for both the analysis of expert text and precise communication. Therefore, an introductory curriculum should pay careful attention to how it incorporates, defines,...
Voltage fault injection is a technique to disrupt power supply, such that the data or instruction flow in a microcontroller can be modified. Recently, a new class of voltage glitches was introduced termed arbitrary wave voltage glitches. Despite its demonstrated success in practical studies it comes with additional challenges, such...
This dissertation delves into understanding, characterizing, and addressing dataset shift in deep learning, a pervasive issue for deployed machine learning systems. Integral aspects of the problem are examined: We start with the use of counterfactual explanations in order to characterize the behavior of deep reinforcement learning agents in visual input...
Compactness in deep learning can be critical to a model’s viability in low-resource applications, and a common approach to extreme model compression is quantization. We consider Iterative Product Quantization (iPQ) with Quant-Noise [Fan et al., 2020] to be state-of-the-art in this area, but this quantization framework suffers from preventable inference...
Secure Computation is a powerful tool that enables a set of parties to jointly compute any function over their private inputs, without a trusted third party. Private Set Intersection is a specific case of two-party Secure Computation, where Alice (with private set X) and Bob (with private set Y) specifically...
The advancement of artificial intelligence (AI) has led to transformative developments across multiple sectors, fostering innovation and redefining our interactions with technology. As AI matures and becomes integrated into society, it offers numerous opportunities to address global challenges and revolutionize a wide array of human endeavors. These advances are driven...
Deep learning is now being utilized widely in applications where sensitive data is being used for model training, for example, in health care. In this scenario, any data leakage will cause privacy concerns to whose data records are used to train the model. An attacker can actively cause privacy leakage...
New capabilities in wireless network security are now possible through deep learning, which can identify and leverage patterns in radio frequency (RF) data. One area of deep learning, known as open set detection, is focused on detecting data instances from new devices encountered during deployment that were not previously seen...
Transmit beamforming is an important technique employed to improve efficiency and signal quality in wireless communication systems by steering signals towards their in- tended users. It often arises jointly with the antenna selection problem due to various reasons, such as limited number of radio frequency (RF) chains and energy/resource effi-...
Shape transformation is a technique for gradually changing one geometric shape to another. A recent approach presents the use of thin-plate radial basis functions as opposed to traditional "blobby sphere" implicit functions. Without the explicit evaluation of he energy function, this approach combined the two traditional steps into one by...
Until now, most hypertext systems have been implemented on large scale computers. With improvements in microprocessors and development of graphical user interfaces, personal computers can run systems that previously needed the power of a mainframe. The low costs and widespread use of PCs will enable many people to use hypertext...
This report addresses the design and implementation of an internet-based grading tool for the "Translators" course. The motivation is to avoid exposing the instructor's Java byte-code to possible reverse-engineering tools and enable students to submit their homework virtually from any machine across the internet. This tool is intended to replace...
Tensor field topology is of importance to research areas of medicine, science, and engineering. Degenerate curves are one of the crucial topological features that provide valuable insights for tensor field visualization. In this thesis, we study the atomic bifurcations of degenerate curves in 3D linear second-order symmetric tensor fields, and...
The core element of democracy is elections. Modern elections not only cost a lot of money to conduct elections, but we also bear a lot of social costs when the election is questioned. For this reason, the US and European countries have been considering ways to innovate by introducing IT...
RNAs play important roles in multiple cellular processes, and many of their functions rely on folding to specific structures. To maintain their functions, secondary structures of RNA homologs are conserved across evolution. These conserved structures provide critical targets for diagnostics and therapeutics. Thus, there is a need for developing fast...
Simultaneous speech translation (SimulST) is widely useful in many cross-lingual communication scenarios, including multinational conferences and international traveling. Since text-based simultaneous machine translation (SimulMT) has achieved great success in recent years. The conventional cascaded approach for SimulST uses a pipeline of streaming ASR followed by simultaneous MT but suffers from...
Alignment of genomic sequences from different species is becoming an increasingly powerful method in biology, and is being used for many purposes. The result of sequence alignments is a list of pairs of matched locations between the pattern string and the text string. However, without any proper visualization tools to...
Learning latent space representations of high-dimensional world states has been at the core of recent rapid growth in reinforcement learning(RL). At the same time, RL algo- rithms have suffered from ignored uncertainties in the predicted estimates of model-free or model-based methods. In our work, we investigate both of these aspects...
Iterative algorithms are simple yet efficient in solving large-scale optimization problems in practice. With a surge in the amount of data in past decades, these methods have become increasingly important in many application areas including matrix/tensor recovery, deep learning, data mining, and reinforcement learning. To optimize or improve iterative algorithms,...
In this thesis, a new learning algorithm is introduced that is targeted towards individual fairness. In order to be individually fair, mispredictions need to be avoided as each such prediction means the learning algorithm was unfair towards some individual. Therefore, achieving individual fairness implies having a perfect classifier, which is...
Many home users nowadays use various smart devices to improve the efficiency and convenience of their home environments. Trigger-action platforms such as “If-This-Then-That” (IFTTT) enable end users to connect different smart devices and services using simple apps to control these devices and automate the tasks (e.g., if the camera detects...
Currently, a popular approach to image classification uses the deep Transformer architecture. In a Transformer, the attention mechanism enables the model to learn efficiently with fewer computational resources than the convolutional neural networks (CNNs). In this thesis, we study the sparse attention mechanism widely used in the Transformers developed specifically...
In standard training regimes, one assumes that the classes presented to a model constitute all of the classes that the model will encounter when it is deployed. In real deployment scenarios, however, a model can sometimes encounter situations or objects that it has never seen. When these scenarios are safety-critical,...
"What’s wrong with this AI?" Explainable AI (XAI) researchers are moving beyond explaining an AI’s actions, to helping users detect an AI’s failures. However this detection may not be enough—for actionability, we often need to pinpoint which part failed. We investigate how AAR/AI, a structured assessment process, supports users with...
Programming is integrated across the workflow of multiple domains where end-user programmers, those who need to program as a means to an end, regularly need to code. In the modern setting of collaborative development, end-user programmers have to interpret the intentions behind existing code to contribute and build solutions to...
With continuing improvements in performance and capability, GPU processing has gained significant and growing interest across science and industry. With this interest, research has increasingly focused upon methods of processing algorithms with stochastic, non-uniform branching while maintaining low divergence. Central among these methods is thread-data remapping (TDR), whereby data is...
Papers proposing novel machine learning algorithms tend to present the algorithm or technique in question in the best possible light. The standard practice is generally for authors to emphasize their proposed algorithms' performance in the precise setting where it is maximally impressive, often by only fully evaluating their best known...
Object detection models are being widely used in many applications, such as autonomous driving, construction management, and cancer detection. Evaluating the performance of the object detection model is more complicated than other computer vision models such as image classification models. Most of the images have several objects to be detected,...
Machine Translation, the task of automatically translating between human languages has been studied for decades. This task is used to be solved by count-based statistical models, e.g. Phrase-based Statistical Machine Translation (PBSMT), which solves the translation problem by separately training a statistical language model and a translation model. Recently, Neural...
The use of board games for teaching introductory computer science is a promising recent avenue of research. The goal is to introduce computing concepts through their use in the implementations of simple games, thereby keeping students engaged through their learning process. However, there is a gap between students' algorithmic descriptions...
Tracr is a modern browser-based user interface, designed to be used with languages that can generate customized explanations from execution traces. While Tracr is primarily designed for use with the Xtra language, Tracr defines a generalized interface that would allow it to be used with other languages as well. Explanations...
This study compares three approaches in the design of an autonomous machine listening agent that predicts harbor porpoise ultrasonic echolocation clicks in diverse noise environments. Considering the temporal variations of noisy coastal ocean soundscapes which the harbor porpoises inhabit, we propose a leave-one-day-out (LODO) cross-validation strategy in the training of...
We explore the application of deep learning to the disparate fields of natural language processing and computational biology. Both the sentences uttered by humans as well as the RNA and protein sequences found within the cells of their bodies can be considered formal languages in computer science, as sets of...
As one of the most popular data types, the point cloud is widely used in various appli- cations, including computer vision, computer graphics and robotics. The capability to directly measure 3D point clouds is invaluable in those applications as depth information could remove a lot of the segmentation ambiguities in...