About Me
CV
Teaching
Research & Talks

Fred Haochen Song

Last updated: January 2026

Education

University of Toronto — Ph.D. in Statistical Science (Bandit Algorithm Personalization and Inference)
Sept 2022 – Present

Johns Hopkins University — M.S.E. in Applied Mathematics & Statistics
Jan 2020 – May 2021

University of Waterloo — Bachelor of Mathematics in Statistics and Actuarial Science
Sept 2014 – Aug 2018

Continued Education

  • Statistical Rethinking (Bayesian course in R and Stan), taught by Richard McElreath (Max Planck Institute for Evolutionary Anthropology)
    Jan 2023 – Feb 2023

Teaching and Learning Design Professional Development

  • Uncommon Sense Teaching Specialization (Coursera): Neuroscientific and cognitive foundations of learning science; practical active learning strategies
    Nov 2024

  • LearnLab Summer School: Design for Active Learning; Cognitive Task Analysis in education; foundations of Learning Engineering; Intelligent Tutoring Systems (Carnegie Mellon University, USA)
    July 2023

  • Teaching Tech Together: Evidence-based teaching for technical topics; learner personas; assessments; cognitive load (Dr. Greg Wilson)
    2024

Teaching

Course Instructor, University of Toronto
Jul 2025 – Aug 2025

  • Refined and delivered a better-paced course compared to the previous summer, rethinking logistics for evaluation, teaching strategies, and approaches to building an active learning environment.
  • Built interactive LearnR modules for students to learn concepts and practice coding outside of class.
  • Maintained a discussion-based and question-friendly classroom environment that encouraged questions and participation.
  • Received outstanding teaching evaluations: 4.8/5.0 overall course quality, 4.9/5.0 classroom enthusiasm.

Course Instructor, University of Toronto
Jul 2024 – Aug 2024

  • Delivered an introductory statistics course with over 250 students and organized course modules using revealJS.
  • Created interactive, coding-based in-class activities to help students engage with lecture content.
  • Maintained a discussion-based classroom environment to encourage participation.
  • Achieved a teaching evaluation score of 4.6/5.0 in overall teaching quality, etc.

Course Instructor and Teaching Fellow (Part-Time), PATH-GEC Academy
Oct 2021 – Sep 2023

  • Taught or co-taught 5 undergraduate and master-level courses, including Statistics, Actuarial and Financial Science, Introduction to Machine Learning, and Data Analytics.
  • Designed course content collaboratively with professors from different universities, supervised teaching assistants, and maintained a small, discussion-focused classroom environment.
  • Guided student groups in each course to apply learned concepts to concrete projects, providing individual feedback and suggestions for improvement.

Research Project Highlights

Adaptive Text Message-Based System for Mental Health Interventions — Project Leader
Apr 2023 – Present
Led the development of an adaptive text-based dialogue system targeting North American young adults, in collaboration with Microsoft, Northwestern University, and Mental Health America, aimed at providing mental health interventions.

  • Developed a targeted dialogue system that has attracted 5,500+ users.
  • Maintained and enhanced the “MOOCLet” platform for treatment configuration and model management.
  • Collaborated across disciplines to analyze data and improve the effectiveness of mental health interventions.

Personalized Physical Activity Recommendations with Bandits and AI — Project Leader
Aug 2023 – Present
Designed a personalized messaging system combining multi-armed bandit algorithms and large language models (LLMs) to motivate physical activity based on user traits and feedback.

  • Conducted an online survey to collect data for training LLMs and to gain insights into user preferences for behavior change concepts.
  • Collected user feedback on message acceptance, personalization perception, and ideal message preferences.
  • Targeted users with high sedentary behavior, focusing on behavior change concepts such as Behavioral Activation, Gratitude, Help Seeking, and Self-Compassion.

Enhancing Personalization in Clinical Trials with Weighted Allocation Probability in Thompson Sampling (WAPTS) — Principal Author
Sept 2023 – Present
Developed WAPTS, a novel algorithm to enhance personalization in finite-horizon clinical trials by reweighting allocation probability in Contextual Thompson Sampling.

  • Refined Contextual Thompson Sampling into WAPTS, improving personalization in finite-horizon settings.
  • Deployed adaptive surveys using WAPTS to evaluate how students’ self-identified familiarity with course content affects end-of-module outcomes.
  • Demonstrated WAPTS’ superiority over traditional methods through comprehensive performance analysis.

A Novel Hypothesis Test Using Allocation Probability for Inference — Principal Author
Oct 2022 – Present
Developed the “Allocation Probability Test,” a new test statistic for adaptively controlled data, with continuous and stability-controlled versions for better performance.

  • Analyzed challenges in hypothesis testing within traditional bandit frameworks.
  • Demonstrated higher power, controlled false positive rate, and faster computation of the test.

Algorithm for Adaptive Experiments Balancing Statistical Analysis with Reward — Co-author
Jul 2022 – Present
Developed a bandit algorithm to balance reward and statistical power in adaptively collected data, focused on mental health interventions.

  • Performed theoretical analysis on expected loss to ensure optimal performance.
  • Deployed the algorithm in a mental health intervention text messaging system.
  • Evaluating data to assess the algorithm’s impact on treatment effectiveness.

Work Experience

Beijing Urban Oasis Property Management Ltd. Co — Strategy Analyst, Department of Enterprise Development
May 2019 – Dec 2019

  • Managed and optimized data sources to create a new database for market dynamics and investable projects in the finance department.
  • Analyzed operation models of target enterprises and conducted feasibility analyses on ongoing projects.
  • Delivered strategic reports and visual presentations; identified trends and provided data-driven recommendations to assist decision-making.
  • Developed a scalable index system to enhance data retrieval and support business analysis.
  • Reduced operation time by 10% by automating processes with VBA programming while ensuring 100% data entry accuracy.

AIBANK (Joint Venture by CITIC Bank and Baidu) — Data Analyst & HRBP, Big Data Department
Feb 2019 – May 2019

  • Supported data processing, user portfolio analysis, and security improvements for banking operations.
  • Processed and analyzed data using HiveQL and SQL, focusing on user portfolio creation and detecting fraudulent online transactions.
  • Visualized and reported key performance indicators (KPIs) to ensure timely decision-making.
  • Collaborated with security team to assess data quality, check for breaches, and improve data security protocols.

Bank of China | Headquarters — Data Analyst, E-finance Department
Sep 2018 – Feb 2019

  • Contributed to data-driven solutions and emerging technology research for financial applications.
  • Researched and implemented anti-cashing techniques for credit card transactions using deep learning and ensemble learning methods.
  • Enhanced the big data system by transforming rule-based models into Bayesian models for user classification and profiling.
  • Developed and maintained Big Data course content, focusing on technologies like big data, cloud computing, blockchain, and artificial intelligence.
  • Represented the bank in a China Merchants Bank competition, predicting ten-year bond interest rates with Python-based machine learning models.

Working Papers

  1. Haochen Song*, D. Hofer*, R. Islambouli, L. Hawkins, A. Bhattacharjee, M. Franklin, and J. Williams (2025).
    Investigating Relationship Between Physical Activity Level with Personalized Behavior Change Message: Connecting Contextual Bandit with Large Language Models.
    3rd Workshop on Causal Inference and Machine Learning in Practice, KDD 2025 (Accepted).

  2. Haochen Song, I. Musabirov, A. Bhattacharjee, A. Durand, M. Franklin, A. Rafferty, and J. Williams (2025).
    Adaptive Experiments Under High-Dimensional and Data Sparse Settings: Applications for Educational Platforms.
    arXiv preprint: arXiv:2501.03999
    Link: https://arxiv.org/abs/2501.03999

  3. Li, T., Nogas, J., Song, Haochen, H. Kumar, A. Durand, A. Rafferty, … & J. Williams (2021).
    Algorithms for adaptive experiments that balance statistical analysis and reward.
    arXiv:2112.08507.

Publications

  1. Song, Haochen, Li, T., Franklin, M., Deliu, N., & Williams, J. (2024).
    Hypothesis Testing in Adaptively Randomized Experiments: Using the Allocation Probabilities for Inference.
    In Proceedings of the Joint Statistical Meeting (JSM) 2024, Portland.

  2. Ilya Musabirov, Mohi Reza, Haochen Song, Steven Moore, Pan Chen, Harsh Kumar, Tong Li, John Stamper, Norman Bier, Anna Rafferty, Thomas Price, Nina Deliu, Audrey Durand, Michael Liut, and Joseph Jay Williams (2025).
    Platform-based Adaptive Experimental Research in Education: Lessons Learned from Digital Learning Challenge.
    Accepted to the International Conference on Learning Analytics and Knowledge (LAK 2025).

  3. Bhattacharjee, A., Song, Haochen, Wu, X., Tomlinson, J., Reza, M., Chowdhury, A.E., Deliu, N., Price, T.W., and Williams, J.J. (2023).
    Informing Users about Data Imputation: Exploring the Design Space for Dealing With Non-Responses.
    Proceedings of the AAAI Conference on Human Computation and Crowdsourcing, 11(1), 14–26.
    DOI: https://doi.org/10.1609/hcomp.v11i1.27544

  4. Song, Haochen & Wen, Beatrix. (2022).
    A Survey of Convolutional Neural Network and Its Variants.
    pp. 37–45. DOI: https://doi.org/10.1145/3556223.3556229

  5. Li, Peiyan, Liang, Xu, & Song, Haochen. (2022).
    A Survey on Implicit Bias of Gradient Descent.
    pp. 108–114. DOI: https://doi.org/10.1109/ICCRD54409.2022.9730384

Conference Presentations

  1. Haochen Song, I. Musabirov, A. Bhattacharjee, A. Rafferty, A. Durand, M. Franklin, and J. Williams (2024).
    WAPTS: A Weighted Allocation Probability Adjusted Thompson Sampling Algorithm for High-Dimensional and Sparse Experiment Settings.
    Conference on Digital Experimentation (CODE) @ MIT, 2024.

  2. Song, Haochen, Li, T., Franklin, M., Deliu, N., & Williams, J. (2024).
    Hypothesis Testing in Adaptively Randomized Experiments: Using the Allocation Probabilities for Inference.
    Presented at the Theory and Foundations of Bayesian Statistics session, JSM 2024, Portland.

  3. Haochen Song, I. Musabirov, B. Wang, P. Chen, A. Bhattacharjee, A. Durand, A. Rafferty, and J. Williams (2024).
    WAPTS: An Allocation Probability Adjusted Thompson Sampling Algorithm for Learner Sourcing.
    5th Workshop on A/B Testing and Platform Enabled Learning Research, Learning @ Scale, Atlanta.

  4. Haochen Song, I. Musabirov, A. Bhattacharjee, A. Rafferty, A. Durand, M. Franklin, and J. Williams (2024).
    WAPTS: A Weighted Allocation Probability Adjusted Thompson Sampling Algorithm for High-Dimensional and Sparse Experiment Settings.
    Conference on Digital Experimentation (CODE) @ MIT, 2023.

Service

  1. Committee Member, University of Toronto Statistics Graduate Student Research Day 2025
  2. Invited Panelist, Workshop on TA Training Leading Your First Tutorial (2025)
  3. Committee Member, University of Toronto Statistics Graduate Student Research Day 2024
  4. Invited Panelist, Graduate Orientation Day 2024
  5. Invited Panelist, Undergraduate Orientation Day 2023
  6. Undergraduate Mentor, Intelligent Adaptive Intervention Lab (2022 – Present)

Skills

  • Software Development for Machine Learning and Experimental Systems: Python, Django, FastAPI, Celery, PostgreSQL (building and managing scalable systems)
  • Data Analysis and Statistical Modeling: R, Stan, MATLAB, PyMC (data analysis and advanced statistical modeling)
  • Actuarial Sciences Associates candidancy: SOA Certificate of Exam P, FM, and SRM; 3 VEE courses

Relevant Courses

Optimal Experiment Design; Analysis of Time Series; Machine Learning Theories; Statistical Probability Theory; Bayesian Optimization; Applied Statistics

Languages

  • English (native)
  • Chinese (native)