Browse Articles

Recent advances in generative AI have made it increasingly hard to distinguish real images from AI-generated ones. Traditional detection models using CNNs or U-net architectures lack precision because they overlook key spatial and frequency domain details. This study introduced a hybrid model combining Convolutional Neural Networks (CNN) with Fast Fourier Transform (FFT) to better capture subtle edge and texture patterns.

Long hospital stays can be stressful for the patient for many reasons. We hypothesized that age would be the greatest predictor of hospital stay among patients who underwent orthopedic surgery. Through our models, we found that severity of illness was indeed the highest factor that contributed to determining patient length of stay. The other two factors that followed were the facility that the patient was staying in and the type of procedure that they underwent.

Here, the authors used machine learning to analyze microscopic images of hair, quantifying various features to distinguish individuals, even within families where traditional DNA analysis is limited. The Discriminant Analysis (DA) model achieved the highest accuracy (88.89%) in identifying individuals, demonstrating its potential to improve the reliability of hair evidence in forensic investigations.

Here, beginning from an initial interest in the possibility to use a computer to automatically solve a geometry diagram parser, the authors developed their own Fast Geometry Diagram Parser (FastGDP) that uses clustering and corner information. They compared their own methods to a more widely available, method, GeoSolver, finding their own to be an order of magnitude faster in most cases that they considered.

The purpose of this study was to determine the necessity of previous non-algorithmic attacks (Adversarial Training) in light of algorithmic defense methods (Gradient Masking and Defensive Distillation) against FGSM attacks. We found a significant increase in image classification accuracy from defense methods with the non-algorithmic defense method compared to ones without. By analyzing the significance with a McNemar test, we determined that the inclusion of non-algorithmic defense methods is still necessary in light of new algorithmic defense methods.

Debates around legacy preferences, recruited athletes, and affirmative action in U.S. college admissions often focus on the belief that graduating from a more selective institution leads to higher future earnings. The study hypothesized a positive correlation between college selectivity and future income due to enhanced resources and opportunities.

People with Type One diabetes often rely on Continuous Blood Glucose Monitors (CGMs) to track their blood glucose and manage their condition. Researchers are now working to help people with Type One diabetes more easily monitor their health by developing models that will future blood glucose levels based on CGM readings. Jalla and Ghanta tackle this issue by exploring the use of AI models to forecast blood glucose levels with CGM data.

A bottleneck in deleting algal blooms is that current data section is manual and is reactionary to an existing algal bloom. These authors made a custom-designed Seek and Destroy Algal Mitigation System (SDAMS) that detects harmful algal blooms at earlier time points with astonishing accuracy, and can instantaneously suppress the pre-bloom algal population.

Authors examine the effectiveness of Large Language Models (LLMs) like BERT, MathBERT, and OpenAI GPT-3.5 in assisting middle school students with math word problems, particularly following the decline in math performance post-COVID-19.

Here, in an effort to develop a model to predict future groundwater levels, the authors tested a tree-based automated artificial intelligence (AI) model against other methods. Through their analysis they found that groundwater levels in Texas aquifers are down significantly, and found that tree-based AI models most accurately predicted future levels.