top of page

Wednesday ACT Student Speakers - 9/24/25

This project focuses on a seemingly simple question: How Square is a Square? The goal is to develop a non-AI scoring algorithm to quantitatively assess “squareness” of shapes. Given an arbitrary polygon, the algorithm assigns a “squarality” value from 0 to 1 (with 1 being “square”) along with defensible justification. There are many applications for such a shape scoring algorithm: manufacturing (ensuring quality standards), robotics (automating industrial sorting/packaging, mapping navigation routes, etc), government planning (congressional district maps), and property analysis/regulation (ensuring house plots are as “square” as possible). Sometimes, AI is used to detect shapes, but for many applications, AI is inappropriate. In addition to practical considerations (e.g. where training is not possible and/or simpler hardware is required), a non-AI based solution is most important in cases where the algorithm cannot be a black box and the answer must be defensible. The project attempted multiple non-AI based algorithms to assess the “squarality” of a shape: a Path-based Approach (”I can be converted to a square in N moves”) and a Best-fit Approach (”The square I am most similar to is X% similar to me”). After exploring both approaches, the Best-fit approach was selected. An algorithm was developed for computing the “distance” between a shape and a specific square, and then a separate process for selecting squares with which to compare the given shape. This algorithm was developed into a web-based application (available at https://squaretest.app) and tested against thousands of shapes and human intuition. The algorithm was able to successfully classify the vast majority of shapes, and to defend its reasoning for the score. The algorithm is also reversible, and can successfully generate shapes given target “squarality” scores.

Riya Mehrotra

How Square is a Square?

This project explores non-AI based shape detection algorithms and creates an algorithm that, given an arbitrary polygon, assigns a "squarality" value from 0 to 1 with a defensible justification.

Read More

Neil Theodore Santos

Eating Away at Cancers

Exploring how adding more vitamins in your diet can potentially reduce the effects of cancers.

Read More
Alcohol use disorder (AUD) is a chronic disease affecting 400 million people worldwide. Epigenetic changes—genomic changes caused by behavioral or environmental factors—influence AUD by affecting a person’s alcohol tolerance and susceptibility to relapse. To better understand these changes, we investigated a protein called Enok, which is essential for olfactory learning and memory and thus may play key roles in addiction. By assessing alcohol sensitivity, tolerance development, and relapse-like behaviors in fruit flies (Drosophila melanogaster), our project tests the hypothesis that deleting the Enok-encoding gene decreases tolerance development, which is associated with alcohol dependence and physiological impairment. Overall, our results suggest that Enok is associated with higher alcohol sensitivity during initial alcohol exposure, but does not cause significant gains or losses in tolerance during relapse. This demonstrates Enok’s potential as a target for addiction, while providing insight into the molecular mechanisms underlying AUD and improving our understanding on why some people are more likely to relapse than others. These findings may also provide data for future epigenetic-based therapies and personalized treatments for AUD. Such breakthroughs have the power to transform treatment for alcohol addiction, giving millions of people a renewed chance at recovery.

Joanne Lin

Rewriting the Script: The Epigenetics of Alcoholism

We investigated how the protein Enok influences genes involved in processes related to alcoholism, by testing alcohol sensitivity, tolerance, and relapse-like behaviors in fruit flies exposed to alcohol.

Read More

© 2025 LAHS STEAM Week

  • Instagram
  • Facebook
  • YouTube
bottom of page