arXiv research

Combines adversarial and interventional robustness for machine learning models.

The paper tackles matching a desired mean in causal systems through shift interventions.

Study counterfactuals in cyclic systems with shifts and scales.

New framework infers causal shifts in event sequences under out-of-domain interventions.

Proposes DRIG for robust predictions using noise interventions.

This work shifts focus from prediction to intervention in social systems.

Bayesian ARMA model with directional shifts captures structural breaks in compositional time series.

Paper shows fairness and domain adaptation can work together.

In foreign exchange markets monotonic rate changes can be observed in time scale of order of an hour on the days that governmental interventions took place. We estimate the starting time of an intervention using this characteristic behavior of the exchange rates. We find that big amount of interventions can shift the a…

We propose a simple method to learn linear causal cyclic models in the presence of latent variables. The method relies on equilibrium data of the model recorded under a specific kind of interventions ("shift interventions"). The location and strength of these interventions do not have to be known and can be estimated f…

We identify which latent factors change between environments in linear causal models.

ADIGen: Automatic, Debiased, and Invariant Counterfactual Generation

BREEDS benchmarks assess model robustness to subpopulation shifts.

While much attention has been given to the problem of estimating the effect of discrete interventions from observational data, relatively little work has been done in the setting of continuous-valued interventions, such as treatments associated with a dosage parameter. In this paper, we tackle this problem by building …

New method uses limited labeled data and multiple starts to adapt models across domains.

IUPM monitors machine learning models under gradual shifts using optimal transport and active labeling.

New method explains ML performance gaps without causal knowledge.

Modern longitudinal studies collect feature data at many timepoints, often of the same order of sample size. Such studies are typically affected by {dropout} and positivity violations. We tackle these problems by generalizing effects of recent incremental interventions (which shift propensity scores rather than set tre…

Study on forecasting methods and their causal implications.

Generative Intervention Models predict perturbation effects without knowing the underlying mechanisms.

TCR simplifies complex models into interpretable causal factors.

New datasets improve fairness research by revealing UCI Adult's limitations.

New algorithm mitigates misspecification amplification in regression models with covariate shift.

Proposes MSS to identify causal structure from heterogeneous environments.

SAMS-VAE models cellular perturbations using sparse additive mechanisms.

SCTL scales causal domain adaptation without prior knowledge.

Neural network learns causal graph structure from data.

Undersampling often outperforms other methods in nonparametric classification.

Shifts in environment between development and deployment cause classical supervised learning to produce models that fail to generalize well to new target distributions. Recently, many solutions which find invariant predictive distributions have been developed. Among these, graph-based approaches do not require data fro…

Method learns diverse models to resist spurious correlations.

Detects harmful distribution shifts in deployed models without false alarms.

Unified Bayesian model explains in-context learning and activation steering in LLMs.

As machine learning is increasingly used to inform consequential decision-making (e.g., pre-trial bail and loan approval), it becomes important to explain how the system arrived at its decision, and also suggest actions to achieve a favorable decision. Counterfactual explanations -- "how the world would have (had) to b…

This paper tackles sequential distribution shifts in representation learning.

AI needs causal inference to avoid being just a correlation machine.

Proposes a method for generating prediction intervals in dose-response models using conformal prediction.

Gradient-based methods can be biased by distributional asymmetries in bivariate categorical data.

MIP framework improves urban flow prediction by adapting to distribution shifts.

Estimates outcomes under hypothetical scenarios using a flexible framework.

Current OOD benchmarks overestimate model robustness to spurious correlations.

Estimates joint causal effects using single-variable interventions on nonlinear models.

NeuralCSA uses neural networks to analyze causal effects under unobserved confounding.

The ability to learn disentangled representations that split underlying sources of variation in high dimensional, unstructured data is important for data efficient and robust use of neural networks. While various approaches aiming towards this goal have been proposed in recent times, a commonly accepted definition and …

A new method learns causal structure from data using amortized inference.

This paper tackles CRL for multi-node interventions, achieving identifiability guarantees.

This work tackles causal graph discovery with stochastic interventions to minimize the number of interventions.

Causal diagrams based on do intervention are useful tools to formalize, process and understand causal relationship among variables. However, the do intervention has controversial interpretation of causal questions for non-manipulable variables, and it also lacks the power to check the conditions related to counterfactu…

Our goal is to identify beneficial interventions from observational data. We consider interventions that are narrowly focused (impacting few covariates) and may be tailored to each individual or globally enacted over a population. For applications where harmful intervention is drastically worse than proposing no change…