arXiv subjects

Yechan Park

Publications and source records attributed to Yechan Park.

10 recordsLinked to original sources

arXiv2026

GS-VLA: Plug-and-Play Viewpoint Canonicalization for Frozen VLA Policies via Gaussian Splatting

This paper proposes a lightweight, plug-and-play framework that improves robustness to viewpoint shifts in Vision-Language-Action (VLA) policies without policy retraining. To our knowledge, this is the first approach to directly leverage 3D Gaussian-based novel-view synthesis for observation-space adaptation in VLA policies. Current VLA performance relies on the implicit assumption that training and deployment camera configurations are identical. Our experiments show that even a small displacement of the camera mount can reduce the success rate on the LIBERO benchmark from about 90% to about 10% in the worst case. Prior approaches, such as large-scale fine-tuning or generative data augmentation, are computationally expensive and risk catastrophic forgetting. To address this, viewpoint shifts are reformulated as a localized novel-view synthesis problem. Under a Locality assumption, that camera perturbations remain within a small bounded region relative to the workspace, viewpoint normalization reduces to a scene- and policy-independent disocclusion task. Our work implements this idea with a 4M-parameter 3D-Gaussian canonicalizer prepended to a frozen VLA policy. Without modifying policy weights, GS-VLA improves performance across three orthogonal axes: (1) Policy architectures, (2) Unseen task suites, and (3) Perturbation scales. These results show that a lightweight visual module can recover a large fraction of the performance lost under viewpoint shift, without policy retraining.

cs.CV↗

arXiv2026

Dynamically Consistent Statistical Decisions

A large literature in econometrics proposes decision rules with optimality guarantees based on ex ante criteria, such as minimax regret. We develop a framework for analyzing the dynamic consistency of such rules and show that, in many empirically relevant settings, the researcher may wish to deviate from the interim prescription of ex ante optimal rules after observing the data realization. To address this problem, we propose and axiomatize two classes of optimality criteria that yield dynamically consistent decision rules.

econ.EM↗

arXiv2026

A Design-Based Approach to Testing and Inference in (Quasi-)Experiments with Spillovers

Economic policies rarely affect only their direct targets. To study these spillovers, researchers summarize who else was treated with a simple exposure measure, such as the share of treated neighbors within a radius. But for many settings, economic theory provides little guidance on choosing the functional form (e.g., ring) of that measure or its parameters (e.g., radius). We show that the data can inform both choices. Correctly specified exposure measures imply orthogonality conditions that can be used for both estimation and testing. We establish consistency and asymptotic normality of the resulting estimator under spatial and network dependence in a design-based framework, with all randomness arising from treatment assignment. We then characterize the efficient moment conditions. Applied to two large-scale anti-poverty programs, the framework supports some prior radius estimates but rejects others. In the latter case, the revised radius yields substantively different policy-effect estimates.

econ.EM↗

arXiv2026

Choosing A Headline Estimand from Matching, DID, and Hybrid Designs: A Minimax-Regret Approach

Researchers using panel data to estimate causal effects routinely choose among three approaches to using past outcomes: difference-in-differences (DID), conditioning on lagged outcomes (matching, M), and a hybrid that does both (DIDM). The corresponding identifying assumptions are non-nested, leaving little guidance on which to report. We give conditions under which the corresponding estimands are ordered, with DIDM bracketed between matching and DID. This makes DIDM the minimax-regret choice among the three under a broad class of loss functions. We recommend reporting DIDM as the headline estimate, with matching and DID as bounds. We illustrate in applications.

econ.EM↗

arXiv2026

Decomposition of Spillover Effects Under Misspecification: Pseudo-true Estimands and a Local-Global Extension

To measure spillovers, researchers often summarize who else was treated with a simple measure, such as the share of treated neighbors. We study what the researcher estimates when that summary is misspecified. We show that the researcher estimates the best approximation to the true policy effect among all functionals of the chosen summary, and that the usual direct and spillover estimates are exactly the components of this approximation. Under a monotonicity restriction, the estimates also preserve the signs of the true effects. We then specialize this framework to a setting common in applications such as cash transfers: treatment spills over both globally, through market equilibrium, and locally, through network externalities, while the researcher models only one channel. The network estimator still recovers the network spillover, and the equilibrium estimator the equilibrium spillover, even when the other channel is ignored. We illustrate with a simulation calibrated to a large cash-transfer experiment.

econ.EM↗

arXiv2025

RE-TRIP : Reflectivity Instance Augmented Triangle Descriptor for 3D Place Recognition

While most people associate LiDAR primarily with its ability to measure distances and provide geometric information about the environment (via point clouds), LiDAR also captures additional data, including reflectivity or intensity values. Unfortunately, when LiDAR is applied to Place Recognition (PR) in mobile robotics, most previous works on LiDAR-based PR rely only on geometric measurements, neglecting the additional reflectivity information that LiDAR provides. In this paper, we propose a novel descriptor for 3D PR, named RE-TRIP (REflectivity-instance augmented TRIangle descriPtor). This new descriptor leverages both geometric measurements and reflectivity to enhance robustness in challenging scenarios such as geometric degeneracy, high geometric similarity, and the presence of dynamic objects. To implement RE-TRIP in real-world applications, we further propose (1) a keypoint extraction method, (2) a key instance segmentation method, (3) a RE-TRIP matching method, and (4) a reflectivity-combined loop verification method. Finally, we conduct a series of experiments to demonstrate the effectiveness of RE-TRIP. Applied to public datasets (i.e., HELIPR, FusionPortable) containing diverse scenarios such as long corridors, bridges, large-scale urban areas, and highly dynamic environments -- our experimental results show that the proposed method outperforms existing state-of-the-art methods in terms of Scan Context, Intensity Scan Context, and STD.

cs.CV↗

arXiv2024

Matching $\leq$ Hybrid $\leq$ Difference in Differences

Since LaLonde's (1986) seminal paper, there has been ongoing interest in estimating treatment effects using pre- and post-intervention data. Scholars have traditionally used experimental benchmarks to evaluate the accuracy of alternative econometric methods, including Matching, Difference-in-Differences (DID), and their hybrid forms (e.g., Heckman et al., 1998b; Dehejia and Wahba, 2002; Smith and Todd, 2005). We revisit these methodologies in the evaluation of job training and educational programs using four datasets (LaLonde, 1986; Heckman et al., 1998a; Smith and Todd, 2005; Chetty et al., 2014a; Athey et al., 2020), and show that the inequality relationship, Matching $\leq$ Hybrid $\leq$ DID, appears as a consistent norm, rather than a mere coincidence. We provide a formal theoretical justification for this puzzling phenomenon under plausible conditions such as negative selection, by generalizing the classical bracketing (Angrist and Pischke, 2009, Section 5). Consequently, when treatments are expected to be non-negative, DID tends to provide optimistic estimates, while Matching offers more conservative ones.

econ.EM↗

arXiv2024

The Informativeness of Combined Experimental and Observational Data under Dynamic Selection

This paper addresses the challenge of estimating the Average Treatment Effect on the Treated Survivors (ATETS; Vikstrom et al., 2018) in the absence of long-term experimental data, utilizing available long-term observational data instead. We establish two theoretical results. First, it is impossible to obtain informative bounds for the ATETS with no model restriction and no auxiliary data. Second, to overturn this negative result, we explore as a promising avenue the recent econometric developments in combining experimental and observational data (e.g., Athey et al., 2020, 2019); we indeed find that exploiting short-term experimental data can be informative without imposing classical model restrictions. Furthermore, building on Chesher and Rosen (2017), we explore how to systematically derive sharp identification bounds, exploiting both the novel data-combination principles and classical model restrictions. Applying the proposed method, we explore what can be learned about the long-run effects of job training programs on employment without long-term experimental data.

econ.EM↗

arXiv2024

A Bracketing Relationship for Long-Term Policy Evaluation with Combined Experimental and Observational Data

Combining short-term experimental data with observational data enables credible long-term policy evaluation. The literature offers two key but non-nested assumptions, namely the latent unconfoundedness (LU; Athey et al., 2020) and equi-confounding bias (ECB; Ghassami et al., 2022) conditions, to correct observational selection. Committing to the wrong assumption leads to biased estimation. To mitigate such risks, we provide a novel bracketing relationship (cf. Angrist and Pischke, 2009) repurposed for the setting with data combination: the LU-based estimand and the ECB-based estimand serve as the lower and upper bounds, respectively, with the true causal effect lying in between if either assumption holds. For researchers further seeking point estimates, our Lalonde-style exercise suggests the conservatively more robust LU-based lower bounds align closely with the hold-out experimental estimates for educational policy evaluation. We investigate the economic substantives of these findings through the lens of a nonparametric class of selection mechanisms and sensitivity analysis. We uncover as key the sub-martingale property and sufficient-statistics role (Chetty, 2009) of the potential outcomes of student test scores (Chetty et al., 2011, 2014).

econ.EM↗

arXiv2023

Asymptotic expansion for batched bandits

In bandit algorithms, the randomly time-varying adaptive experimental design makes it difficult to apply traditional limit theorems to off-policy evaluation of the treatment effect. Moreover, the normal approximation by the central limit theorem becomes unsatisfactory for lack of information due to the small sample size of the inferior arm. To resolve this issue, we introduce a backwards asymptotic expansion method and prove the validity of this scheme based on the partial mixing, that was originally introduced for the expansion of the distribution of a functional of a jump-diffusion process in a random environment. The theory is generalized in this paper to incorporate the backward propagation of random functions in the bandit algorithm. Besides the analytical validation, the simulation studies also support the new method. Our formulation is general and applicable to nonlinearly parametrized differentiable statistical models having an adaptive design.

stat.ME↗