Shannon Entropy Research Tool for Analyzing DNA/Amino Acid Sequences and Solar Eruptions Data with Application to Generative AI Diffusion Model of Text To Image Technology
Year: 2025
Type: conference-paper
Abstract: The Shannon entropy formula has been accepted as a calculation to measure the amount of information in a random trial.The recent success of Generative AI diffusion model of text to image technology offers much inspiration for STEM students to learn about random processes in research projects beyond the regular classes in a community college setting.The 1-dimensional DNA/amino acid sequences on PubMed serve as signal examples of information content analysis using Shannon entropy calculations.The extension to 2-dimensional image entropy calculations could use the Solar Dynamics Observatory data as practice examples.The computation experience learned on the PubMed and SDO data analysis components was found to be sufficient to start an analysis of using entropy calculation to classify the output of the diffusion model used in Generative AI.The research projects using PubMed data (such as the ZENK gene involved in songbird singing) and SDO data (such as the NOAA 13664 driven 2024 Mother's Day Solar Storm) yielded satisfactory results and fulfilled the criterion of authentic experience in undergraduate student research.For the Generative AI diffusion model component, the entropy classification of the forward diffusion was found to be acceptable, while the backward diffusion remains as work-in-progress.Future research topics using entropy calculation are discussed, together with sustainability.
Authors: Julianne Torreno, Nealesh Guha, Mashtura Rahman, Michael Ventouratos, David Lee, Shivansh Sharma, Sunil Dehipawala, Guozhen An, Tak Cheung
Institution: Queensborough Community College, CUNY
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4409905274