Speaker: Silvia Lara
Title: Low-light image enhancement
Abstract: Vision systems heavily rely on the presence of sufficient light and low levels of noise. Images of night scenes, however, do not follow these assumptions. In the literature there are two main algorithms to enhance and denoise dark images, one based on fully connected neural networks and one based on illumination maps. We combine the strengths of each model to create a more effective denoising and enhancing algorithm using Deep Learning. Our model consists of a convolutional autoencoder network with skip connections. In this talk, we will explore the process of developing the new model through hypotheses creation and verification. We will set up a process to test the model’s capabilities and limitations, including in the denoising and enhancing aspect, and the models’ prediction quality in low vs. high definition images.
Speaker: Alex Meyer
Title: The Borders Are Moving: A Gerrymandering App for the Average American
Abstract: This talk focuses on building a web application for the layperson about gerrymandering. It first provides context on gerrymandering, gerrymandering measures, and the current sites and tools available to the public. It then describes how features are chosen for the application and the implementation of those features. Finally, it summarizes user feedback and the iterations of the application.