Four Drexel College of Computing & Informatics (CCI) students were selected to present at the Pennoni Honors College’s 2022 Week of Undergraduate Excellence virtual poster session (May 16 to 20).
The Week of Undergraduate Excellence seeks to highlight and celebrate undergraduate excellence and accomplishments in research, scholarship, and creative work from across the University. All undergraduate students are invited and encouraged to present academic/creative work throughout the week-long celebration and to attend events that showcase student excellence.
The following CCI students presented during the virtual poster session:
Maria Minodora Mares (BS computer science) - "AI and The Skinner Box Experiment" (Faculty Mentor: Brian Stuart, PhD)
- View Poster Presentation Site
- View poster image
- Abstract: Exploring the topic of artificial intelligence indicates that academics are interested in learning, which is one of the most intrinsically intelligent behaviors. The Skinner Box experiment investigates instrumental learning by watching an animal's learning path to a reward (typically food) in a controlled setting (a cage with a bar at one end). For this research project, Maria used Dr. Brian Stuart's Cybernetic Automaton, an adaptive finite state machine with probabilistic outputs, to study how it behaves in the previously mentioned experiment and finally assess its potential to execute instrumental learning. They were able to make the world more realistic by adding extra directions and locations to the basic setup (which was extremely simplistic – two places, four directions). The model's degree of learning was consistent throughout the numerous adjustments they made to the environment, and it corresponded with Skinner Box-related findings in psychological literature.
Satvik Tripathi (BS computer science) - "Turing Test Inspired Method for Analysis of Biases Prevalent in Artificial Intelligence Based-Radiology" (Faculty Mentor: Edward Kim, PhD)
- View Poster Presentation Site
- View poster image
- Abstract: Because of the rising need to deliver better global healthcare, the development of computer-based and robotic healthcare technology based on artificial intelligence has risen. One such innovation is the Turing test. The Turing test was developed to measure artificial intelligence (AI) in computer technology. It is still an important qualitative tool for assessing the next generation of medical diagnostics and medical robots. Satvik used a meta-analytical assessment framework to evaluate Turing test outcomes using quantitative approaches. He created a test environment for the suggested methodology's real execution. This modified Turing Test is a particularly powerful benchmark for measuring the real performance of the AI model on a range of edge cases and typical instances, as well as determining if the algorithm is biased towards any one type of case. Not only can we discover biases, but we can also classify the type of bias and attempt to resolve it.
Ryu Morgan (BS software engineering) – “Using MURAL to Facilitate Collaboration in a Virtual Design Session” (Faculty Mentor: Ellen Bass, PhD)
Supported as part of the Pennoni Honors College’s Pathway to SuperNova program
- View Poster Presentation Site
- View poster image
- Abstract: Before the COVID-19 pandemic human factors engineers would have face-to-face design meetings with subject matter experts (SMEs) to create and execute work system design modifications. However, the COVID-19 outbreak necessitated a shift to a virtual environment. The focus of this "Pathway to SuperNova" study was on a method for facilitating design sessions in a virtual environment. This project entailed investigating digital whiteboard collaborative areas and determining applicable application functions, such as various color-coded sticky notes. The collaboration program MURAL was discovered to allow participants to view structured whiteboards while researchers assigned to data input may contribute theme data in the background. MURAL delivered real-time color-coded, scaled, and editable digital sticky notes. The collaboration program MURAL was discovered to allow participants to view structured whiteboards while researchers assigned to data input may contribute theme data in the background. MURAL delivered real-time color-coded, scaled, and editable digital sticky notes. This project also included the creation of a technique for carrying out the design session with SMEs, human factors engineers (i.e., facilitators), notetakers, and Mural editors. The participants would not have to learn the new technology if Mural editors submit data on an organized Mural board. The final step was to have a virtual design session with health care SMEs. Mural editors played an important part in the design process without the health care SMEs having to learn how to utilize a new digital whiteboard collaboration tool.
Ahn Oliver Nguyen (BS computer science) – “Supporting Comparison of Process Map Data for Collaborative Teams” (Faculty Mentor: Ellen Bass, PhD)
Supported as part of the “Handoffs and Transitions in Critical Care - Understanding Scalability” project (PI Meghan Lane-Fall, MD, MSHP, University of Pennsylvania)
- View Poster Presentation Site
- View poster image
- Abstract: Human-factor engineers must be capable of comparing various team-based processes. These processes are represented visually in the form of Process Maps. These processes, comprising of Phases, Tasks, and Roles, may also be represented as structured data. Data comparison in either format is error prone. This project is dedicated to facilitating the comparison of Process Map structured data. A tool should either compare two sets of process map data or compare one set of process map data with input on discrepancies in execution by separate teams to assist the analysts' workflow. Iterative design and development were used to create the Parser Tool. Requirements and use cases were iteratively developed with the analysts who were manually creating and comparing the Process Maps. The JavaScript-based parser can validate and compare pairs of process map data using spreadsheets and text documents as inputs. The output distinguishes between the two procedures as well as faults in the input files.