Traffic's Toll on Health and Innovating Anomaly Detection Data

Bad Data Generation for Anomaly Detection by Dr. Marcella Astrid

Anomaly detection is integral to numerous domains such as surveillance, medical diagnostics, security systems, and industrial processes. However, the rarity and diverse characteristics of anomalies present substantial hurdles in acquiring ample training data. In this presentation, we investigate methodologies for generating low-quality data to supplement existing datasets, thereby enhancing anomaly detection capabilities. Specifically, we delve into the types of low-quality data that can yield maximum benefits for model performance. Moreover, we underscore the broader significance of data augmentation beyond binary classification tasks, illustrating its relevance in practical applications.

How does a 1% increase in traffic cost your health? by Dzidas Martinaitis

The Luxembourg government provides open access to data on a variety of subjects, including economic indicators, health, transportation, and environmental measurements. In this research project, I utilized a dataset on pollution to precisely quantify the impact of traffic on NO2 emissions. Using causal inference techniques such as regression analysis, Bayesian inference, and Double Machine Learning, this work delivers precise estimates in response to the main question - how much a pollution cost our health? The approach shown in the presentation can serve as a model for answering important questions in both business and academic contexts.