Search NASASearch

Engineering topics

Yoon, Hong-Jun

Publications and source records attributed to Yoon, Hong-Jun.

Leveraging hyperspectral imaging to identify drought tolerant Populus species and genotypes within species

The aim of this study was to identity variation in drought tolerance across genotypes of Populus deltoides, Populus trichocarpa, and hybrids of the two species. A panel of 102 Populus genotypes, comprising 37 genotypes of P. trichocarpa, 37 of P. deltoides and 28 unique hybrid genotypes (P. trichocarpa x P. deltoides and P. deltoides x P. trichocarpa) were evaluated in the greenhouse under two treatments, well-watered (WW) and drought (DS). Plant physiological data were collected throughout the experiment once the drought treatment began. Throughout the experiment, we tracked soil volumetric water content, pot weight, stomatal conductance, quantum yield of photosystem II, and electron transport rate. In addition to those measurements, upon completion of the experiment, we assessed above and belowground plant biomass, plant height and stem diameter, leaf number, specific leaf area, relative water content, total protein, and total chlorophyll. We obtained hyperspectral signatures of one leaf from each plant at the end of the experiment. Columns BC – LL are hyperspectral averages for one leaf from each plant at each wavelength as described in the column header.

Hyper-spectral imaging, Populus, plant stress tole

Improving Text Classification with Large Language Model-Based Data Augmentation

Large Language Models (LLMs) such as ChatGPT possess advanced capabilities in understanding and generating text. These capabilities enable ChatGPT to create text based on specific instructions, which can serve as augmented data for text classification tasks. Previous studies have approached data augmentation (DA) by either rewriting the existing dataset with ChatGPT or generating entirely new data from scratch. However, it is unclear which method is better without comparing their effectiveness. This study investigates the application of both methods to two datasets: a general-topic dataset (Reuters news data) and a domain-specific dataset (Mitigation dataset). Our findings indicate that: 1. ChatGPT generated new data consistently enhanced model’s classification results for both datasets. 2. Generating new data generally outperforms rewriting existing data, though crafting the prompts carefully is crucial to extract the most valuable information from ChatGPT, particularly for domain-specific data. 3. The augmentation data size affects the effectiveness of DA; however, we observed a plateau after incorporating 10 samples. 4. Combining the rewritten sample with new generated sample can potentially further improve the model’s performance.

97 MATHEMATICS AND COMPUTING