Publications

Causal effect estimation from observational data typically assumes a single fixed observational distribution. We study the setting in which the distribution is known only to belong to a credal set, a convex set of plausible observational laws. Our results characterize how this observational ambiguity is transformed by causal identification formulas. First, using standard two-point information-theoretic tools after composition with the interventional map, we derive lower bounds for causal effect estimation. The key message is a causal amplification phenomenon: observational discrepancies in low-propensity treatment strata can be hard to detect while producing separated interventional effects. Consequently, in the binary hard-pair constructions, sample complexity scales as Ω(1/(\pi_0\varepsilon^2)) or Ω(1/(\beta_0\varepsilon^2)), and the average treatment effect need not incur an additional dependence on the number of strata. Second, we prove Lipschitz bounds for the interventional mapping, with amplification factor (1+1/\alpha_\min), showing how positivity controls robust identification width. Third, we study minimax regret for causal decisions under credal uncertainty and show that randomization can halve worst-case regret in a symmetric action-separation construction. We also give matching upper bounds in special cases and an NP-hardness result for computing robust identification width. The experiments are diagnostic checks of the tight hard-instance scaling laws rather than broad empirical benchmarks.

The Rashomon effect, the fact that many different models can achieve nearly identical predictive accuracy, has been extensively studied across various domains of machine learning. While modern conformal prediction provides valid intervals for a single model, it remains blind to the inherent arbitrariness of having to select from multiple candidates. This paper bridges this gap by investigating when model multiplicity, or the existence of many near-optimal models, materially widens prediction intervals. We prove a sharp phase transition for ridge regression under Gaussian design and introduce *Rashomon Prediction Intervals* (RPIs), which bound the range of predictions across the Rashomon set and establish a Rashomon-Conformal Bridge that provides distribution-free coverage guarantees. A critical tolerance \varepsilon^⋆= Θ(σ^2/p) separates two distinct regimes: below it, Rashomon-conformal intervals match standard conformal width; above it, model multiplicity dominates, and this scaling is provably minimax optimal. Computationally, exact RPIs are polynomial for convex classes but NP-hard for tree ensembles, with a thin-cap volumetric barrier ruling out sampling-based approximation. Experiments on six UCI benchmarks confirm the phase transition, with all datasets falling in the sub-critical regime at standard tolerances, indicating that model multiplicity is invisible in prediction intervals for typical tabular regression.

A text2sql agent for Vietnamese corporation financial report data.

Detecting Vietnamese names from a global name collection is essential for connecting human resources within the Vietnamese community worldwide and for developing Vietnamese Expert Finding Systems. However, applying standard classification methods to recognize Vietnamese names is challenging due to their ambiguity with names from other countries. In this study, we proposed our novel approach, Highly Discriminative N-grams (HDN), which utilizes statistics from n-grams from human names for the classification task. We show that HDN is a lightweight and simple-to-reproduce approach for this task by benchmarking with other machine learning models.

Using engagement data from 253 undergraduates on VnCodelab, the Stacking Classifier outperformed five individual machine learning models in predicting student grade ranges, highlighting the potential of data-driven forecasting for early intervention and more informed educational assessment.

We evaluate World Action Models and Vision-Language-Action model in action space and feature space.