Christos Faloutsos Fredkin University Professor of Computer Science Website CMU Scholars Page ORCiD Office 7003 Gates and Hillman Centers Email christos@andrew.cmu.edu Phone (412) 268-1457 Department Computer Science Department Administrative Support Person Diana Hyde Research Interests Artificial Intelligence Machine Learning Systems Databases Distributed Systems Advisees Catalina Vajiac Saranya Vijayakumar CSD Courses Taught 15826 - Fall, 2025 15826 - Fall, 2024 There are two main focus areas: graph mining and stream mining. In the first, the goal is to find patterns in large graphs, so that we can spot anomalies, communities, patterns and regularities. Graphs appear in many instances: as document-term bipartitegraphs in Information retrieval, as web pages or blogs linking to each other, as customer-product recommendations, as protein-protein regulatory networks, as computer-network traffic, and many more. Our emphasis is on scalability, so that we can handle graphs withthousands and millions of nodes. Research directions include time-evolving graphs, where we have beenusing 'tensors' to find patterns, as well as graphs where the nodes and/or the edges have attributes. The second research area focuses on streams, which are semi-infinitenumerical time series. The setting also has numerous applications, like sensor data monitoring, motion capture data, automatic alerts in the 'self-*' PetaByte storage system, chlorine level monitoring on the drinking water, and several more. The emphasis is to develop algorithms that inspect every measurementonly once, and then discard it, since we can not affort to store the huge volume of historical data. The common threads in both areas are the power-laws and the existenceof self-similarity. Real graphs have skewed, Zipf-like degree distributions, and consist of communities-within-communities. Similarly, real sensor measurements are often bursty, but still self-similar, with bursts within bursts. We use or develop tools that exactly exploit the power laws and self-similarity, to find better patterns and anomalies than standard tools would find. keywords: Database Management Systems, Data Mining, Graphs, Social Networks, Network Security. Publications Preprint CurvGAD: Leveraging Curvature for Enhanced Graph Anomaly Detection 2025 Grover K, Gordon GJ, Faloutsos C Conference Featpilot: Automatic Feature Augmentation on Tabular Data 2025 • Proceedings - International Conference on Data Engineering • 00:2148-2160 Liang J, Lei C, Qin X, Zhang J, Katsifodimos A, Faloutsos C, Rangwala H Conference HybGRAG: Hybrid Retrieval-Augmented Generation on Textual and Relational Knowledge Bases 2025 879-893 Lee M-C, Zhu Q, Mavromatis C, Han Z, Adeshina S, Ioannidis VN, Rangwala H, Faloutsos C Conference Kronecker Generative Models for Power-Law Patterns in Real-World Hypergraphs 2025 • PROCEEDINGS OF THE ACM WEB CONFERENCE 2025, WWW 2025 • 1261-1272 Choe M, Ko J, Kwon T, Shin K, Faloutsos C Preprint LAKEGEN: A LLM-based Tabular Corpus Generator for Evaluating Dataset Discovery in Data Lakes 2025 Dai Z, Lei C, Katsifodimos A, Qin X, Faloutsos C, Rangwala H
Preprint CurvGAD: Leveraging Curvature for Enhanced Graph Anomaly Detection 2025 Grover K, Gordon GJ, Faloutsos C
Conference Featpilot: Automatic Feature Augmentation on Tabular Data 2025 • Proceedings - International Conference on Data Engineering • 00:2148-2160 Liang J, Lei C, Qin X, Zhang J, Katsifodimos A, Faloutsos C, Rangwala H
Conference HybGRAG: Hybrid Retrieval-Augmented Generation on Textual and Relational Knowledge Bases 2025 879-893 Lee M-C, Zhu Q, Mavromatis C, Han Z, Adeshina S, Ioannidis VN, Rangwala H, Faloutsos C
Conference Kronecker Generative Models for Power-Law Patterns in Real-World Hypergraphs 2025 • PROCEEDINGS OF THE ACM WEB CONFERENCE 2025, WWW 2025 • 1261-1272 Choe M, Ko J, Kwon T, Shin K, Faloutsos C
Preprint LAKEGEN: A LLM-based Tabular Corpus Generator for Evaluating Dataset Discovery in Data Lakes 2025 Dai Z, Lei C, Katsifodimos A, Qin X, Faloutsos C, Rangwala H
