Estimating counterfactual treatment outcomes over time in complex multiagent scenarios
Authors:
Keisuke Fujii,
Koh Takeuchi,
Atsushi Kuribayashi,
Naoya Takeishi,
Yoshinobu Kawahara,
Kazuya Takeda
Abstract:
Evaluation of intervention in a multiagent system, e.g., when humans should intervene in autonomous driving systems and when a player should pass to teammates for a good shot, is challenging in various engineering and scientific fields. Estimating the individual treatment effect (ITE) using counterfactual long-term prediction is practical to evaluate such interventions. However, most of the conven…
▽ More
Evaluation of intervention in a multiagent system, e.g., when humans should intervene in autonomous driving systems and when a player should pass to teammates for a good shot, is challenging in various engineering and scientific fields. Estimating the individual treatment effect (ITE) using counterfactual long-term prediction is practical to evaluate such interventions. However, most of the conventional frameworks did not consider the time-varying complex structure of multiagent relationships and covariate counterfactual prediction. This may lead to erroneous assessments of ITE and difficulty in interpretation. Here we propose an interpretable, counterfactual recurrent network in multiagent systems to estimate the effect of the intervention. Our model leverages graph variational recurrent neural networks and theory-based computation with domain knowledge for the ITE estimation framework based on long-term prediction of multiagent covariates and outcomes, which can confirm the circumstances under which the intervention is effective. On simulated models of an automated vehicle and biological agents with time-varying confounders, we show that our methods achieved lower estimation errors in counterfactual covariates and the most effective treatment timing than the baselines. Furthermore, using real basketball data, our methods performed realistic counterfactual predictions and evaluated the counterfactual passes in shot scenarios.
△ Less
Submitted 17 February, 2024; v1 submitted 4 June, 2022;
originally announced June 2022.
Systematic characterization of upper critical fields for MgB$_2$ thin films using the two-band superconducting theory
Authors:
S. Noguchi,
A. Kuribayashi,
T. Ishida,
T. Oba,
H. Iriuda,
M. Yoshizawa,
Y. Harada,
S. Miki,
H. Shimakage,
Z. Wang,
K. Satoh,
T. Yotsuya
Abstract:
We present experimental results of the upper critical fields $H_{\rm c2}$ of various MgB$_2$ thin films prepared by the molecular beam epitaxy, multiple-targets sputtering, and co-evaporation deposition apparatus. Experimental data of the $H_{\rm c2}(T)$ are successfully analyzed by applying the Gurevich theory of dirty two-band superconductivity in the case of $D_π/D_σ>1$, where $D_π$ and…
▽ More
We present experimental results of the upper critical fields $H_{\rm c2}$ of various MgB$_2$ thin films prepared by the molecular beam epitaxy, multiple-targets sputtering, and co-evaporation deposition apparatus. Experimental data of the $H_{\rm c2}(T)$ are successfully analyzed by applying the Gurevich theory of dirty two-band superconductivity in the case of $D_π/D_σ>1$, where $D_π$ and $D_σ$ are the intraband electron diffusivities for $π$ and $σ$ bands, respectively. We find that the parameters obtained from the analysis are strongly correlated to the superconducting transition temperature $T_{\rm c}$ of the films. We also discuss the anormalous narrowing of the transition width at intermediate temperatures confirmed by the magnetoresistance measurements.
△ Less
Submitted 8 November, 2008;
originally announced November 2008.
