Organizational Unit:

Undergraduate Research Opportunities Program

Permanent Link

https://hdl.handle.net/1853/70802

Parent Organization

Organizational Unit

Office of Undergraduate Education

ArchiveSpace Name Record

https://finding-aids.library.gatech.edu/agents/corporate_entities/1180

Full item page

Publication Search Results

Now showing 1 - 10 of 94

Multidimensional Allocation: In Apportionment and Bin Packing

(Georgia Institute of Technology, 2022-08) Chiu, Alvin

In this thesis, we deal with two problems on multidimensional allocation, specifically in apportionment and in bin packing. The apportionment problem models the allocation of seats in a House of Representatives such that it is proportional to the dimensions being represented. One example is the allocation of the 435 House seats to the 50 U.S. states, which demands being proportional in the one dimension of state population. It is also common to demand proportionality in both state population and political affiliation, where we now have to allocate to two dimensions simultaneously. We begin by investigating what it means for an 1-D apportionment to be "fair", and use this to judge the various methods of apportionment that have been used throughout history. This motivates the study of divisor methods, a certain class of apportionment methods that avoid any paradoxes. We then formally tackle the problem of 1-dimensional and 2-dimensional apportionment with divisor methods through the lens of optimization. The optimization approach generalizes well to higher dimensions, but a proportional apportionment is not always possible in 3 or more dimensions. Our thesis outlines the current method for finding "approximate" apportionments and improves it in certain regimes. As for bin packing, we model the allocation of virtual machines to servers (with limited capacity) in cloud computing, with the goal of designing and analyzing efficient algorithms that optimize the expected cost of the allocation. This builds off previous work that only considered the case where the items being packed had a one-dimensional size. We extend some of those results to items with multi-dimensional size in this thesis.
Six Feet Apart: Online Payments During the COVID-19 Pandemic

(Georgia Institute of Technology, 2022-05) Shaikh, Omar

Since the COVID-19 pandemic, businesses have faced unprecedented challenges when trying to remain open. Because COVID-19 spreads through aerosolized droplets, businesses were forced to distance their services; in some cases, distancing may have involved moving business services online. In this work, we explore digitization strategies used by small businesses that remained open during the pandemic, and survey/interview small businesses owners to understand preliminary challenges associated with moving online. Furthermore, we analyze payments from 400K businesses across Japan, Australia, United States, Great Britain, and Canada. Following initial government interventions, we observe (at minimum for each country) a 47% increase in digitizing businesses compared to pre-pandemic levels, with about 80% of surveyed businesses digitizing in under a week. From both our quantitative models and our surveys/interviews, we find that businesses rapidly digitized at the start of the pandemic in preparation of future uncertainty. We also conduct a case-study of initial digitization in the United States, examining finer relationships between specific government interventions, business sectors, political orientation, and resulting digitization shifts. Finally, we discuss the implications of rapid & widespread digitization for small businesses in the context of usability challenges and interpersonal interactions, while highlighting potential shifts in pre-existing social norms.
Examinator: Detecting Exam Cheating Via Comparison of Question Answering Timings

(Georgia Institute of Technology, 2022-05) Sonubi, Imanuel Oluseyi

Cheating is an issue that affects more than just the student doing it, no matter the format of the assessment being cheated on. Take-home exams provide more flexibility for the instructor and student than regular proctored exams, but it is that lack of proctoring during the exam that makes cheating trickier to detect -- students may meet up outside of the classroom and inappropriately collaborate on these tests even though they are to be done individually. Examinator aims to detect cheating on Canvas take-home exams by examining the times at which students view questions and comparing them with other students' times to find any exam attempts that have suspiciously similar timestamps for each question.
Learning Rotation-in-Place and Orbiting Policies for a Quadruped Robot

(Georgia Institute of Technology, 2022-05) Kim, Alex

Reinforcement learning (RL) algorithms have successfully learned control policies for quadruped locomotion such as walking, rotation, and basic navigation. We utilize Proximal Policy Optimization and iGibson to train a quadruped robot in simulation to do two specific tasks—rotation-in-place and orbiting—with orbiting being a novel, previously unexplored task for quadruped robots. We show that with proper reward and environment engineering, we are able to train a simple two layer fully-connected neural network to do both tasks. We propose that both policies will be useful in a larger control system for a quadruped robot to explore its environment and that the orbiting policy is both novel and useful for learning more about certain objects of interest in the environment. See policy video here: https://youtu.be/olk2hJ372a4.
CopyCat: Leveraging American Sign Language Recognition in Educational Games for Deaf Children

(Georgia Institute of Technology, 2022-05) Ravi, Prerna

Deaf children born to hearing parents lack continuous access to language, leading to weaker working memory compared to hearing children and deaf children born to deaf parents. CopyCat is a game where children communicate with the computer via American Sign Language (ASL), and it has been shown to improve language skills and working memory. Previously, CopyCat depended on unscalable hardware such as custom gloves for sign verification, but modern 4K cameras and pose estimators present new opportunities. This thesis focuses on the current version of the CopyCat game using off-the-shelf hardware, as well as the state-of-the-art sign language recognition system we have developed to augment game play. Using Hidden Markov Models (HMMs), user independent word accuracies were 90.6%, 90.5%, and 90.4% for AlphaPose, Kinect, and MediaPipe, respectively. Transformers, a state- of-the-art model in natural language processing, performed 17.0% worse on average. Given these results, we believe our current HMM-based recognizer can be successfully adapted to verify children’s signing while playing CopyCat.
NeuroMapper: Using UMAP to visualize the training dynamic of neural network

(Georgia Institute of Technology, 2022-05) Zhou, Zhiyan

Deep neural networks are often considered as black box models. One of the reasons is that they have a large number of parameters. Understanding the model through these parameters is often a hard task. Recent advances in nonlinear dimensionality reduction techniques offer a better way to visualize high-dimensional nonlinear data[1]. These techniques have the potential of helping people understand deep neural networks. However, many of these techniques pose problems, such as indeterministic results and slower computation. These problems prevent them from being used in visualizing the training dynamic of deep neural networks[2]. Neuro-Mapper attempts to tackle this through the use of scalable data visualization technologies and Aligned UMAP, a special kind of UMAP that is able to generate embeddings that are aligned with each other[3]. Users can explore the training behavior of a model using NeuroMapper and adjust the hyperparameter as they see fit. NeuroMapper works across multiple platforms and can visualize more than 40000 data points.
Artificial Intelligence (AI) through Symbiosis

(Georgia Institute of Technology, 2022-05) Ponda, Devansh Jatin

While most traditional machine learning approaches tend to focus more on data – cleaning, filtering, and formatting data – I ask if we could instead focus more of our attention on trying to make our computers understand the nuances of the real- world instead. What if instead of humans feeding cleaned data to computers, there could be a ‘symbiosis’ in which the computer actively learns the habits of humans through their everyday interactions with the world, and assist them in these interactions? In my research, I have tried to build one of the earliest systems that follow the AI- through-Symbiosis framework in the context of order picking. By collecting video data from a head-mounted GoPro camera that a user wears while picking items, I try to recognize – without any data labelling – which item the user has picked. The idea is that if this information is known, we can then use it to determine if the user has picked an incorrect item and alert them, which would in turn help reduce the error rates in the process. In this thesis, we explore the idea of AI through Symbiosis, the motivation to study it, the methodology of collecting data in the context of order picking, preliminary results, and possible future work.
ASL Fingerspelling Recognition Through Hidden Markov Models

(Georgia Institute of Technology, 2022-05) So, Matthew

American Sign Language (ASL) is notable for its unique grammatical structures such as classifiers and its use in a more complex multi-dimensional (spatial) medium, rather than a uni-dimensional (auditory) medium. As a result, most current recognition or translation systems require expensive equipment such as accelerometers and depth cameras, restrictions on recognizable phrases, and/or low recognition accuracy. However, ASL fingerspelling representing signs corresponding to the letters of the alphabet, a subset of ASL, may be able to engage and encourage parents for initial learning. In this project, we have analyzed data sufficiency by collecting training data from three users on 36 coarticulated fingerspelled words of varying lengths using cameras recording hand at a size of approximately 256 by 256 pixels. We have also tested model recognition accuracy on training data using uniletter and triletter Gaussian Mixture-Hidden Markov Models (GMM-HMMs) and using various levels of data augmentation involving including finger features (locations) centered around the location of the hand and including wrist features. We have found that triletter GMM-HMMs produce up to 96% letter and word accuracy rates on user-dependent tests when centered hand features and wrist features are included in training data. The results indicate that fingerspelling recognition on mobile devices is viable and should be further investigated.
Improving the Balance of a Random-access Data Structure on a Migratory Thread Platform

(Georgia Institute of Technology, 2022-05) Joshipura, Muktaka B.

In this work, we look at a distributed random-access data structure provided with a migratory-thread platform: the Lucata Pathfinder, called a chunked array. A chunked array provides sequential locality on sections of the array while having its memory distributed. We consider drawbacks of the implementation of this structure in terms of balancing the distribution of its elements. We provide an alternate implementation with better balancing, which can simplify algorithms like a distributed sort. We show analytically that our implementation balances significantly better for smaller numbers of elements, but the improvement declines proportionally to the number of elements. Finally, we test our implementation in a simulator and show that the balancing of the elements also leads to good balancing of execution.
Linear Promises: Towards Safer Concurrent Programming

(Georgia Institute of Technology, 2022-05) Rau, Ohad

In this paper, we introduce a new type system based on linear typing, and show how it can be incorporated in a concurrent programming language to track ownership of promises. By tracking write operations on each promise, the language is able to guarantee exactly one write operation is ever performed on any given promise. This language thus precludes a number of common bugs found in promise-based programs, such as failing to write to a promise and writing to the same promise multiple times. We also present an implementation of the language, complete with an efficient type checking algorithm and high-level programming constructs. This language serves as a safer platform for writing high-level concurrent code.