A single link to the first track to allow the export script to build the search page
  • Undergraduate Poster Abstracts
  • Computer/Systems Engineering

    FRI-721 PARALLEL COMPUTING METHODS WITH TEMPORAL ORDER-PRESERVING HASHING FOR GESTURE RECOGNITION

    • Joey Velez-Ginorio ;
    • Guo-Jun Qi ;

    FRI-721

    PARALLEL COMPUTING METHODS WITH TEMPORAL ORDER-PRESERVING HASHING FOR GESTURE RECOGNITION

    Joey Velez-Ginorio, Guo-Jun Qi.

    College of Engineering and Computer Science, University of Central Florida, Orlando, FL.

    The advent of increasingly powerful parallel computing architectures has allowed their increased use within realms of computer vision and machine learning. In general, these architectures grant an increased capability to handle problems unfit for conventional sequential processing techniques. Using this ability to process data in parallel provides the unique opportunity to efficiently handle tasks in image and video processing. Within this scope, we seek to improve the efficiency of a novel hashing algorithm for gesture recognition, specifically through its implementation on a common unified device architecture (CUDA)-based parallel architecture. This involves parallelizing the algorithm so as to split the computational efforts across many CUDA cores, as opposed to one central processing unit. Specific to vectorized computation, this approach yields measurable improvements as a consequence to inherently large iterative constraints within sequential frameworks. In regards to the latter methodology, extra care must is considered so as not to alter key components of the hashing algorithm during parallelization. These key aspects preserve the temporal structure of the dataset, providing scalability and resisting effects of temporal warping, a vulnerability common in similar models. Retaining this necessary design yields is paramount, as initial experiments on public 3D human action datasets unveiled the potential for recognition accuracies of above 80%. Further developments within the parallelized framework look to maintain or improve these accuracies while providing definitive improvements in efficiency. Contingent on our success, the findings provide grounds for an enhancement in a host of applications regarding human-to-computer interaction via recognition of gestures.

    THU-721 SPACE WEATHER MEASUREMENTS USING GLOBAL NAVIGATION SATELLITE SYSTEMS (GNSS)

    • Karielys Ortiz-Rosario ;
    • Wayne Scales ;
    • Marc Jean ;

    THU-721

    SPACE WEATHER MEASUREMENTS USING GLOBAL NAVIGATION SATELLITE SYSTEMS (GNSS)

    Karielys Ortiz-Rosario2, Wayne Scales1, Marc Jean1.

    1Virginia Polytechnic Institute and State University, Blacksburg, VA, 2Escuela de Ingeneria Jose Domingo Perez, Universidad del Turabo, Gurabo, PR.

    Global navigation satellite systems (GNSS) are now commonly used for geographic positioning, timing data, economic applications (such as international banking), space weather studies, and atmospheric science analysis. GNSS currently consists of 4 constellations: the U.S.'s Global Positioning System (GPS), the Russian Federation's Global Orbiting Navigation Satellite System (GLONASS), the European Union's Galileo, and China's BeiDou Navigation Satellite System (BDS). GNSS signaling can be adversely affected by the concentration of plasma (the collection of free moving electrons and ions) in the Earth’s ionosphere. Therefore, the measurement of the total electron content (TEC) in the ionosphere is particularly essential in order to reduce the data error and generate more accurate results. The objective of this project is to use pseudorange data, the measured distance between the GNSS satellites and their receivers, from all 4 GNSS constellations to study space weather and its impact on navigation systems using TEC calculations. Therefore, live data was collected from NovAtel, Ashtech, and ASTRA GNSS receivers, as well as the NASA Crustal Dynamic Data Information System (CDDIS) website. Subsequently, a MATLAB script was generated to use RINEX3 navigation and observation files to acquire pseudorange data and perform TEC calculations. This work presents the data analysis and shows that the TEC peaked at dusk and dawn. Statistics on the data analysis are used to explain the inaccuracy within the GNSS signaling due to the TEC concentration.

    THU-738 MEASURING THE EFFECTIVENESS OF ETIQUETTE STRATEGIES TO MITIGATE NEGATIVE EMOTIONS

    • Jordan Zonner ;
    • Euijung Yang ;

    THU-738

    MEASURING THE EFFECTIVENESS OF ETIQUETTE STRATEGIES TO MITIGATE NEGATIVE EMOTIONS

    Jordan Zonner1, Euijung Yang2.

    1Doane College, Crete, NE, 2Iowa State University, Ames, IA.

    This research explored the importance of applying human-to-human etiquette strategies to human computer interaction. The effectiveness of etiquette strategies was measured and compared to visualize their importance in the role of learning. In the past decade, the number of people taking online classes has increased. Simultaneously, people are using more technology, which increases the number of human-computer interactions. Etiquette strategies are strategic ways of phrasing communication, which people use to adapt their conversations appropriately. Unlike humans, computers lack the awareness and ability to adapt to these human emotions. For this research, the users were manipulated in order to induce negative emotions and mitigate them through etiquette strategies in the interest of testing their effectiveness. These strategies were evaluated in order to see if presenting the same information with different etiquette styles could positively affect the user’s performance, motivation, and emotional state. The results demonstrated rising trends that certain etiquette strategies can positively affect the user’s learning experience when used to mitigate negative emotions, such as frustration.