Dark Energy Explorers: Utilizing Human Labeling and Machine Learning to Understand Dark Energy 

In this talk, I investigate dark energy through lyman-alpha emitting (LAE) galaxies and baryon acoustic oscillations as cosmic probes via the Hobby-Eberly Telescope Dark Energy Experiment (HETDEX). HETDEX is an unbiased spectroscopic survey and therefore, one significant challenge is labeling the copious amounts of data. I have focused my research on combining artificial intelligence (AI) and human labeled sources. This work began with the creation and launch of the NASA Zooniverse project Dark Energy Explorers in 2021. Participants of the project learn how to classify the LAEs and I have created a community of 23,000 volunteers representing 159 countries. Dark Energy Explorers is available in 9 different languages and growing. The Dark Energy Explorers have made 8 million classifications so far and through this work we can confidently determine if a source is an artifact at 94% confidence level. Each source is classified by at least ten individuals; this confidence level increases for higher signal-to-noise sources. I discovered this as an efficient and effective way to clean datasets, which led to a reliable probability on 1.2 million Lyman-alpha emitting galaxy detection candidates. This unique combination of public engagement with machine learning algorithms, t-SNE and k-nearest neighbor, has proven efficient and effective for cleaning large survey data. Lastly, in this talk, I look to the future of continued work combining these techniques with the Hobby-Eberly Telescope Dark Energy Experiment and the Vera Rubin Observatory while also growing our tools for astronomy education and public trust of AI.