Muhammad Hasan Ferdous

Ph.D. Candidate in Information Systems (defending Summer 2026) · University of Maryland, Baltimore County

I am a Ph.D. candidate in Information Systems at the University of Maryland, Baltimore County, and will defend my dissertation in Summer 2026. My research addresses causal discovery in multivariate time series data exhibiting autocorrelation, non-stationarity, multi-seasonality, and irregular sampling. The work synthesizes statistical inference, machine learning, and information systems, building on a B.Sc. in Statistics from the University of Dhaka and graduate training in the UMBC Causal AI Lab under Dr. Md Osman Gani.

My work has appeared in KDD, MLHC, AAAI, ICMLA, and IEEE PerCom Workshops, with two manuscripts currently under review at TMLR and the KDD 2026 Dataset Track. The methodological contributions include CDANs and eCDANs, constraint-based discovery algorithms with optimized conditioning sets for autocorrelated and non-stationary time series; DCD, a decomposition-based framework for multi-seasonal data; and the TimeGraph benchmark suite for evaluating temporal causal discovery methods. The CDANs algorithm is publicly available as a Python package: pip install cdans. I have applied these methods through collaborations with climate scientists at the University of Colorado Boulder on Arctic sea ice prediction and Greenland supraglacial lake evolution, and with the UMBC Causal AI Lab on healthcare informatics.

I have served as a Graduate Teaching Assistant at UMBC for five consecutive semesters across four courses: Database Program Development, Advanced Database Project, Management Information Systems, and Structured Systems Analysis and Design. I have also developed a graduate lecture module on Causal AI. My pedagogical approach emphasizes active learning through the Flipped Classroom model and Think-Pair-Share exercises, the integration of authentic research datasets from healthcare and climate science, and frequent low-stakes assessment via embedded quizzes. Inclusive instruction that supports students across varying levels of preparation is central to my practice.

I am on the academic job market for Fall 2026 or Spring 2027 starts and am applying for tenure-track Assistant Professor positions in Computer Science, Information Systems, and Data Science. I welcome inquiries from departments seeking faculty at the intersection of methodological research and undergraduate mentorship. Email: h.ferdous@umbc.edu.

CV (PDF) Research Statement Teaching Statement Diversity Statement

News

  • Summer 2026 Ph.D. defense at UMBC. On the academic job market for Fall 2026 and Spring 2027 starts.
  • May 2026 Awarded the COEIT Research Day 2026 Student Award for the poster "G-DCD: Generalized Decomposition-based Causal Discovery for Multivariate Multi-Seasonal Temporal Data".
  • February 2026 DCD preprint released (arXiv:2602.01433) and submitted to Transactions on Machine Learning Research.
  • 2026 CDANs released as an open-source Python package: pip install cdans (GitHub).
  • December 2025 Causal Time Series Modeling of Supraglacial Lake Evolution in Greenland under Distribution Shift accepted at IEEE ICMLA 2025 (arXiv).
  • October 2025 ClassyGlass: A Benchmark Dataset for Activity and Mobility Analysis using Smart Eyewear submitted to the KDD 2026 Dataset Track.
  • August 2025 Awarded the COEIT Summer Student Project Award ($5,000) from UMBC for student-led research.
  • August 2025 TimeGraph: Synthetic Benchmark Datasets for Robust Time-Series Causal Discovery published at KDD 2025 (paper, DOI).
  • May 2025 Honorable Mention for Research Poster, COEIT Research Day, UMBC.
  • March 2025 Correlation to Causation: A Causal Deep Learning Framework for Arctic Sea Ice Prediction presented at IEEE PerCom Workshops 2025 (DOI).
  • 2023 CDANs: Temporal Causal Discovery from Autocorrelated and Non-Stationary Time Series Data published at MLHC 2023 (New York). MLHC Travel Award.
  • 2023 eCDANs: Efficient Temporal Causal Discovery from Autocorrelated and Non-Stationary Data (Student Abstract) at AAAI 2023.

Featured Publications

  • TimeGraph: Synthetic Benchmark Datasets for Robust Time-Series Causal Discovery
    Muhammad Hasan Ferdous, Emam Hossain, Md Osman Gani
    KDD 2025 · Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining
  • CDANs: Temporal Causal Discovery from Autocorrelated and Non-Stationary Time Series Data
    Muhammad Hasan Ferdous, Uzma Hasan, Md Osman Gani
    MLHC 2023 · Proceedings of the 8th Machine Learning for Healthcare Conference
    Code
  • PreprintDCD: Decomposition-based Causal Discovery from Autocorrelated and Non-Stationary Temporal Data
    Muhammad Hasan Ferdous, Md Osman Gani
    arXiv:2602.01433 · Under review at Transactions on Machine Learning Research
    arXiv

Full publications →

Open-Source Software

Reproducible research and accessible tools are central to my research program. Three software contributions are publicly released: CDANs (pip install cdans) for constraint-based causal discovery, TimeGraph for synthetic benchmarking of temporal causal discovery methods, and ClassyGlass for multimodal wearable-sensor analysis. See the software portfolio →