“I Don’t See Color”: An Analysis of Racial Diversity within Prime Time Television

Overview

Mr. Kenneth W. Roedl Racial representation on television has been changing ever since television was invented. This study aims to do three things: 1.) Track the trends television has been setting since the 1960s and showcase what the future of television might look like; 2.) Identify the audience’s opinions on the growing diversification within television as well as what racial groups believe about the past, present, and future of diversity on television; 3.) Understand the impact that diversity (or the lack of) on television has had on viewers based on their perception on past shows as well as their self-confidence after watching tv. Using the research of Riva Tukachinsky and Dana Mastro, this project aimed to dive more deeply into the presence of people of color on television since the start of television and less on their portrayal. The study looks to answer two questions: “Is diversity within primetime television becoming more mainstream?” and “What are primetime television viewers’ opinions on diversity within primetime television?”. This is done by creating an examination for television shows on primetime television (ABC, NBC, and CBS). The survey also shows how minorities agree that representation is necessary, as opposed to Caucasians, and how they felt underrepresented on television as a child while whites believe they were represented on television as a child and that they are still represented now. The research will potentially serve as a blueprint for further studies to look into race portrayal and the future of racial representation on television which could be extended out to other countries and point out differences in television representation. Counting With Entropy Using information Entropy to Detect Colloidal Particles in Holographic Video Microscopy Images Zoe Rutkovsky The Packer Colegiate Institute Brooklyn, New York Supervising Scientist: David G. Grier New York University Holographic Video uses lasers to visualize and track colloidal particles in solution. The video can track the particles’ 3-dimensional position on a focal plane. Holographic images of microscopic polystyrene spheres provide an assortment of data, including sphere size, refractive index, depth of field, volume and entropy. This data can be used to characterize a large variety of microscopic particles. Prior to this study, the ability to determine the density of a solution was primarily limited to dynamic light scattering and machine learning methods. This research has produced a new technique for determining density, which is far more efficient than Dynamic Light Scattering (DLS). Howe ver, conventional methods of particle imaging and recognition (such as DLS, and Machine Learning) have proven difficult and the data is costly to analyze, often requiring prior training. Using entropy measurements as means of calculating the number of particles per image is proposed as an efficient method. The entropy and particle count can be determined using a Python based code. This code looks at the information content of individual images, and has been successful as it can determine how many images in a data set are blank and possibly determine density. This advance has implications in biopharmaceuticals, semiconductor processing and wastewater management. Predicting Intensity Maps of Cosmic Neutral Hydrogen from Dark Matter Using Convolutional Neural Netw orks Helen Shao The Bronx High School of Science Rego Park, New York Supervising Scientist: Dr. Francisco Villaescusa The Flatiron Institute One of the main goals in cosmology is to constrain the values of the parameters in the ΛCDM model. This is achieved by comparing observational data from cosmological surveys with theoretical models of the universe generated in hydrodynamic simulations. However, the computational cost of these simulations limit the data that scientists have access to. An alternative would be to use machine learning to generate the simulations of a particular detectable constituent of the early universe, neutral cosmic hydrogen. In this paper, I showed that neural network algorithms can be used to accurately and efficiently predict the intensity maps of HI in all regions of the cosmic web given the input of dark matter density field. Specifically, the UNET architecture can generate HI maps with speeds up to a hundred times faster than the traditional method of creating Illustris simulations. Moreover, I found that the statistical properties of the predictions are similar to those in the Illustris maps with the use of validation metrics such as the power spectrum and 1D Probability Distribution Function. This work is an important step towards the goal of obtaining fast and accurate models of our universe that can be used with the upcoming 21cm cosmological surveys. Furthermore, this efficient approach is crucial for the future of cosmology when analysis of the physical properties of the universe is needed on a faster and more accessible scale.

Competition history

• JSHS 2020 Category not listed

Resources

• National JSHS abstract book ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.