Cognitive science research workspace — Active Inference, Bayesian modeling, ant behavior, and computational neuroscience

Hierarchical Bayesian modeling toolkit for interoceptive psychophysics (HRDT & RRST). Includes Stan models, power analysis tools, and educational resources for researchers.

Football forecasting framework to simulate the FIFA World Cup using team strength modeling and probabilistic match prediction.

Agent-based political election simulator using predictive coding

The project focuses on the analysis of experimental data regarding the continuous biomass production of a cyanobacterium of the genus Nostoc. The primary objective is to understand how different cultivation factors influence biological growth over time, monitored through optical density (OD).

Inflation forecasting during crisis periods using Bayesian Dynamic Linear Models, traditional econometrics, and machine learning. Includes data, code, and comprehensive analysis report.

An R package for modeling asymmetric spatial associations between cell types in tissue images using a multilevel Bayesian framework.

Stan implementation of Lee & Sarnecka's (2010, 2011) knower-level model

computational model showing how compulsive urges might arise when the brain stops listening to its own body signals, testing this idea against real physiological data like heart rate and pupil size.

Bayesian uncertainty, calibration, and hallucination-risk modeling for reliable language and multimodal AI.

R code and Stan models for "Decoupling Decision and Intensity in Visual Attention" .

Bayesian Marketing Mix Modeling project using PyMC to estimate marketing channel effectiveness through probabilistic inference, adstock transformation, uncertainty-aware ROI analysis, and seasonal trend modelling across multiple advertising channels.

Portfolio of applied data science projects in Bayesian modeling, causal inference, hierarchical models, and agent-based simulation.

A repository for the R code and data used in Iwasaki et al. (2023)

Probabilistic Graphical Models Project

Adaptive learning classifier using a Beta–Binomial model to estimate student proficiency

Comprehensive analysis of differential gene expression using Bayesian statistics and advanced statistical modeling techniques. The project includes scripts, data, figures, and analysis results.

Three‑level hierarchical enactive inference model of mental action (focused‑attention meditation with expert–novice profiles) with reproducible simulations and figures.

Bayesian media mix modeling (MMX) framework with latent state decomposition and SKAN bias corrections for historical & causal attribution

RL was cheaper. The heuristic was safer. Neither was correct. POLARIS stress-tests operational policies under chaos, demand spikes, and black swan events asking one question: which policy survives when everything goes wrong? Built with constrained RL, Bayesian modeling, CVaR risk metrics, and a human-in-the-loop governor.