ºÚÁϸ£ÀûÍø

Tackling two core problems in data analysis

focused on two fundamental problems in machine learning: K-center clustering, which groups similar data points, and submodular maximization, a general optimization framework that models many decision-making tasks.

Both play a central role in applications such as pattern recognition, resource allocation, and network analysis.

Algorithms for massive data models

Her research explored these problems within three major computational settings that reflect how real-world systems operate: distributed environments, where data is processed across many machines; dynamic models, where information changes over time; and streaming models, where data arrives continuously and memory is limited.

By designing efficient algorithms for these scenarios, Biabani demonstrated how large-scale data can be processed quickly and accurately.

Making clustering more robust

In the first part of her thesis, Biabani developed methods to group similar data points even when the dataset contains errors or irregular entries. This makes it possible to analyze large, irregular datasets more reliably, which is important for applications such as detecting patterns in social media, tracking traffic flows, or monitoring health data.

Her work is among the first to provide strong guarantees that these methods will perform well in real-world, large-scale systems.

Optimizing under change

In the second part of her thesis, Biabani developed methods to make decisions that rely on large datasets more adaptable. Her algorithms can quickly update solutions when new data is added or existing data changes, ensuring that outcomes remain accurate even as conditions evolve.

This is especially useful for applications such as optimizing resources, improving recommendation systems, or monitoring complex networks, where data is constantly changing and decisions need to stay up to date.