![header.png](https://static.wixstatic.com/media/262923_7726920c22754e0da687407ba32df6b9~mv2.png/v1/fill/w_940,h_300,al_c,q_85,enc_auto/header.png)
Team Supermassive Blackhole: Movie Finder
written by
Olivia Haimerl
Thursday, February 16, 2023
Congratulations to Team Supermassive Blackhole on winning The Erdős Institute’s Spring 2022 Data Science Bootcamp with their project Movie Finder!
Comprised of University of Michigan students Anna Brosowsky, Sayantan Khan, Nancy Wang, Ethan Zell, and Yili Zhang, Team Supermassive Blackhole successfully combined three existing data sets to build a movie search app that gives the title of a movie using what the user remembers about the plot. The user can also input the approximate year and genre of the movie to provide additional information for the search engine. For example, inputting “cowboy doll gets jealous of spaceman toy,” provides the search result of the 1995 film Toy Story with an accuracy of 0.84. According to the team, “our model prototype achieves 84% accuracy on the test data, which is a big improvement over the 21% achieved by our baseline model of comparing substring similarity scores.”
When discussing how the team settled upon these data sets and specific project, Ethan noted “there was a need to have a project that could have multiple goals and levels of success so we could be quite ambitious but also ensure that we have a product at the end of the bootcamp.” He emphasized there was a need to have a deliverable that would demonstrate why the project matters. To guarantee that the search engine would be working by the end of the four-week intensive bootcamp, Yili highlighted the importance of teamwork within the group: “we split our tasks so some team members focused their efforts on data cleaning while others worked more on machine learning. It is all about teamwork and a little bit about divide and conquer, so each of us had our own tasks and we would meet regularly throughout the week to discuss.” Anna further noted the tasks were split by interest, familiarity, and practicality. Team Supermassive Blackhole attributes their success to their ability to work together in a group, their ability to complete their own tasks in a timely manner, their interest in the project, and their project mentor, Frank Hidalgo.
The team noted that with more time and access to other resources, they envision a variety of future possibilities for the app. This includes allowing users to continually train the data by providing feedback on the search engine results and expanding the app to include books or tech-related information, such as finding answers to the question: “new iPad update issue with Google Drive.” Sayantan also noted that fellow students outside of the team utilized the search engine in opposition to its intended purpose: to find movies to watch based on arbitrary movie plot inputs.
Although the team agrees their success was due to intensive teamwork, they noted that all future bootcamp participants should aim to start their projects as soon as possible! Further, when addressing the future bootcamp participants, Nancy noted: “individuals should apply to the data science bootcamp as soon as they can, even if they are not sure if they want to go into industry or not, because it is an extremely useful skill and it would help them more when they apply for internships or jobs.”
Congratulations again to Team Supermassive Blackhole as well as all of the other teams who completed a Spring 2022 Data Science Bootcamp project!
TEAM
Advancing Cardiac Diagnostics: Machine Learning Approaches for ECG-Based Heart Condition Analysis and Reconstruction
Gbocho Masato Terasaki
![clear.png](https://static.wixstatic.com/media/55f531_b9f3f13ce3aa4af78af2cc6d3563b81b~mv2.png/v1/fill/w_3,h_3,al_c,lg_1,q_85,enc_auto/clear.png)
The project aims to advance cardiac health diagnostics through two critical tasks: irregular heartbeat classification and activation map reconstruction using ECG signals. Utilizing the ECG Heartbeat Categorization Dataset, the project focuses on developing a robust multiclass classification model to accurately diagnose various irregular heartbeats, facilitating early detection and timely treatment of cardiac conditions. In parallel, the project employs a neural network to transform ECG sequences into activation maps, using simulated intracardiac transmembrane voltage recordings. This reconstruction task is designed to deepen our understanding of the heart’s electrical activity, offering detailed insights that can lead to more precise and effective clinical interventions.
![](https://static.wixstatic.com/media/a994932411404ef3bb797ba005125f5d.png/v1/fill/w_45,h_45,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/a994932411404ef3bb797ba005125f5d.png)
![](https://static.wixstatic.com/media/a994932411404ef3bb797ba005125f5d.png/v1/fill/w_45,h_45,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/a994932411404ef3bb797ba005125f5d.png)
![](https://static.wixstatic.com/media/a994932411404ef3bb797ba005125f5d.png/v1/fill/w_45,h_45,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/a994932411404ef3bb797ba005125f5d.png)