Dhruva Paul

Revamping Movies with AI

Started Nov. 29, 2022

Abstract or project description

Because of the simple recording equipment in the early to mid 20th century, hundreds of incredible movies have been stuck with black-and-white video and patchy, static-like audio. Many have forgotten about these classic films and many more refuse to try and watch these movies because of the poor video and audio. Recent advancements in graphical and audio manipulation through computer science models have made it possible to alter each of these fields. By mapping black and white pixels to corresponding red, green, and blue pixels, one would be able to fully transform black-and-white video to colored video. As for audio, one would be able to isolate certain sounds through signal processing, amplify them, and reduce background static, yielding a much clearer audio in the overall film. Combining these two methods, with the help of research mentors at Polygence, would create a unique model that unites both graphical and audio manipulation for proper refinement of vintage films. With such a model, old movies could be easily converted to more modern formats and would be widely more accessible to viewers.

Neural Network model predictions of monthly changes in Apple stock based on the company’s current and past financial data

Started Apr. 16, 2022

Abstract or project description

Neural Network model predictions of monthly changes in Apple stock based on the company’s current and past financial data

With the rise of increasingly powerful machine learning algorithms, many have applied these concepts to the stock market in the hopes of creating a portfolio with the maximum amount of profit possible. However, most predictors simply rely on data derived from stock prices themselves in order to make these predictions, such as candlestick trends and patterns. The model this project is working on will focus on one company, Apple, and predict its stock with not only previous stock prices of the company, but also the stock prices of the NASDAQ, of which Apple is a part of. The model will also take into account the stock prices of the other major players of the NASDAQ, such as Amazon and Netflix. In addition, the model will input the financial data of the company, including variables such as the company’s revenue, gross profit, as well as other financial metrics such as gross profit margin and return on equity. The global price of materials of Apple’s most bought products will also be accounted for. These series extend from 2017 to 2022, providing 5 years worth of data. The model will utilize a neural network to identify patterns within this data and determine the monthly change in Apple stock. I chose a neural network in order to use a deep learning model to go through hundreds of data points in order to find trends within the data that would take a human ages to uncover.