Introducing Variational Autoencoders to High School Students
Project Overview
The document explores the integration of Variational Autoencoders (VAEs) in K-12 education, emphasizing their role in teaching generative AI through artistic expression. It highlights a lesson plan that utilizes interactive tools and philosophical concepts to engage students in understanding the workings and applications of VAEs, particularly in creative arts. Pilot studies conducted with high school students demonstrated positive outcomes in their comprehension of VAEs, indicating that this innovative approach not only fosters engagement but also enhances AI literacy within educational curricula. Overall, the findings suggest that incorporating generative AI technologies like VAEs can effectively enrich learning experiences and equip students with essential skills for the future.
Key Applications
Variational Autoencoders (V AEs)
Context: High school students learning about AI concepts through creative applications.
Implementation: Developed a lesson plan incorporating a web-based game and Google Colab notebook to teach V AEs using the philosophical metaphor of Plato's cave.
Outcomes: Students understood the roles of the encoder and decoder, engaged with interactive tools, and expressed enjoyment in the hands-on experiences.
Challenges: Students without coding experience struggled with the Google Colab interface and the concepts of encoder, decoder, and latent space.
Implementation Barriers
Technical Barrier
Students with little or no coding experience found the Google Colab interface complex and confusing.
Proposed Solutions: Simplify the interface and provide clearer instructions on using the notebook.
Knowledge Barrier
Students lacked foundational knowledge in AI concepts and neural networks, making it difficult to grasp the roles of V AEs.
Proposed Solutions: Integrate this lesson into a broader Introduction to AI curriculum that covers foundational concepts.
Project Team
Zhuoyue Lyu
Researcher
Safinah Ali
Researcher
Cynthia Breazeal
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Zhuoyue Lyu, Safinah Ali, Cynthia Breazeal
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai