Abstract:Structure-agnostic causal inference studies how well one can estimate a treatment effect given black-box machine learning estimates of nuisance functions (like the impact of confounders on treatment and outcomes). Here, we find that the answer depends in a surprising way on the distribution of the treatment noise. Focusing on the partially linear model of \citet{robinson1988root}, we first show that the widely adopted double machine learning (DML) estimator is minimax rate-optimal for Gaussian treatment noise, resolving an open problem of \citet{mackey2018orthogonal}. Meanwhile, for independent non-Gaussian treatment noise, we show that DML is always suboptimal by constructing new practical procedures with higher-order robustness to nuisance errors. These \emph{ACE} procedures use structure-agnostic cumulant estimators to achieve $r$-th order insensitivity to nuisance errors whenever the $(r+1)$-st treatment cumulant is non-zero. We complement these core results with novel minimax guarantees for binary treatments in the partially linear model. Finally, using synthetic demand estimation experiments, we demonstrate the practical benefits of our higher-order robust estimators.
Abstract:Faithful evaluation of language model capabilities is crucial for deriving actionable insights that can inform model development. However, rigorous causal evaluations in this domain face significant methodological challenges, including complex confounding effects and prohibitive computational costs associated with extensive retraining. To tackle these challenges, we propose a causal representation learning framework wherein observed benchmark performance is modeled as a linear transformation of a few latent capability factors. Crucially, these latent factors are identified as causally interrelated after appropriately controlling for the base model as a common confounder. Applying this approach to a comprehensive dataset encompassing over 1500 models evaluated across six benchmarks from the Open LLM Leaderboard, we identify a concise three-node linear causal structure that reliably explains the observed performance variations. Further interpretation of this causal structure provides substantial scientific insights beyond simple numerical rankings: specifically, we reveal a clear causal direction starting from general problem-solving capabilities, advancing through instruction-following proficiency, and culminating in mathematical reasoning ability. Our results underscore the essential role of carefully controlling base model variations during evaluation, a step critical to accurately uncovering the underlying causal relationships among latent model capabilities.
Abstract:This paper investigates the integration of response time data into human preference learning frameworks for more effective reward model elicitation. While binary preference data has become fundamental in fine-tuning foundation models, generative AI systems, and other large-scale models, the valuable temporal information inherent in user decision-making remains largely unexploited. We propose novel methodologies to incorporate response time information alongside binary choice data, leveraging the Evidence Accumulation Drift Diffusion (EZ) model, under which response time is informative of the preference strength. We develop Neyman-orthogonal loss functions that achieve oracle convergence rates for reward model learning, matching the theoretical optimal rates that would be attained if the expected response times for each query were known a priori. Our theoretical analysis demonstrates that for linear reward functions, conventional preference learning suffers from error rates that scale exponentially with reward magnitude. In contrast, our response time-augmented approach reduces this to polynomial scaling, representing a significant improvement in sample efficiency. We extend these guarantees to non-parametric reward function spaces, establishing convergence properties for more complex, realistic reward models. Our extensive experiments validate our theoretical findings in the context of preference learning over images.
Abstract:We address the problem of estimating heterogeneous treatment effects in panel data, adopting the popular Difference-in-Differences (DiD) framework under the conditional parallel trends assumption. We propose a novel doubly robust meta-learner for the Conditional Average Treatment Effect on the Treated (CATT), reducing the estimation to a convex risk minimization problem involving a set of auxiliary models. Our framework allows for the flexible estimation of the CATT, when conditioning on any subset of variables of interest using generic machine learning. Leveraging Neyman orthogonality, our proposed approach is robust to estimation errors in the auxiliary models. As a generalization to our main result, we develop a meta-learning approach for the estimation of general conditional functionals under covariate shift. We also provide an extension to the instrumented DiD setting with non-compliance. Empirical results demonstrate the superiority of our approach over existing baselines.
Abstract:Performing policy evaluation in education, healthcare and online commerce can be challenging, because it can require waiting substantial amounts of time to observe outcomes over the desired horizon of interest. While offline evaluation methods can be used to estimate the performance of a new decision policy from historical data in some cases, such methods struggle when the new policy involves novel actions or is being run in a new decision process with potentially different dynamics. Here we consider how to estimate the full-horizon value of a new decision policy using only short-horizon data from the new policy, and historical full-horizon data from a different behavior policy. We introduce two new estimators for this setting, including a doubly robust estimator, and provide formal analysis of their properties. Our empirical results on two realistic simulators, of HIV treatment and sepsis treatment, show that our methods can often provide informative estimates of a new decision policy ten times faster than waiting for the full horizon, highlighting that it may be possible to quickly identify if a new decision policy, involving new actions, is better or worse than existing past policies.
Abstract:This paper proposes the automatic Doubly Robust Random Forest (DRRF) algorithm for estimating the conditional expectation of a moment functional in the presence of high-dimensional nuisance functions. DRRF combines the automatic debiasing framework using the Riesz representer (Chernozhukov et al., 2022) with non-parametric, forest-based estimation methods for the conditional moment (Athey et al., 2019; Oprescu et al., 2019). In contrast to existing methods, DRRF does not require prior knowledge of the form of the debiasing term nor impose restrictive parametric or semi-parametric assumptions on the target quantity. Additionally, it is computationally efficient for making predictions at multiple query points and significantly reduces runtime compared to methods such as Orthogonal Random Forest (Oprescu et al., 2019). We establish the consistency and asymptotic normality results of DRRF estimator under general assumptions, allowing for the construction of valid confidence intervals. Through extensive simulations in heterogeneous treatment effect (HTE) estimation, we demonstrate the superior performance of DRRF over benchmark approaches in terms of estimation accuracy, robustness, and computational efficiency.
Abstract:Reinforcement Learning from Human Feedback (RLHF) is widely used to align Language Models (LMs) with human preferences. However, existing approaches often neglect individual user preferences, leading to suboptimal personalization. We present the Preference Pretrained Transformer (PPT), a novel approach for adaptive personalization using online user feedback. PPT leverages the in-context learning capabilities of transformers to dynamically adapt to individual preferences. Our approach consists of two phases: (1) an offline phase where we train a single policy model using a history-dependent loss function, and (2) an online phase where the model adapts to user preferences through in-context learning. We demonstrate PPT's effectiveness in a contextual bandit setting, showing that it achieves personalized adaptation superior to existing methods while significantly reducing the computational costs. Our results suggest the potential of in-context learning for scalable and efficient personalization in large language models.
Abstract:Estimates of causal parameters such as conditional average treatment effects and conditional quantile treatment effects play an important role in real-world decision making. Given this importance, one should ensure these estimators are calibrated. While there is a rich literature on calibrating estimators of non-causal parameters, very few methods have been derived for calibrating estimators of causal parameters, or more generally estimators of quantities involving nuisance parameters. In this work, we provide a general framework for calibrating predictors involving nuisance estimation. We consider a notion of calibration defined with respect to an arbitrary, nuisance-dependent loss $\ell$, under which we say an estimator $\theta$ is calibrated if its predictions cannot be changed on any level set to decrease loss. We prove generic upper bounds on the calibration error of any causal parameter estimate $\theta$ with respect to any loss $\ell$ using a concept called Neyman Orthogonality. Our bounds involve two decoupled terms - one measuring the error in estimating the unknown nuisance parameters, and the other representing the calibration error in a hypothetical world where the learned nuisance estimates were true. We use our bound to analyze the convergence of two sample splitting algorithms for causal calibration. One algorithm, which applies to universally orthogonalizable loss functions, transforms the data into generalized pseudo-outcomes and applies an off-the-shelf calibration procedure. The other algorithm, which applies to conditionally orthogonalizable loss functions, extends the classical uniform mass binning algorithm to include nuisance estimation. Our results are exceedingly general, showing that essentially any existing calibration algorithm can be used in causal settings, with additional loss only arising from errors in nuisance estimation.
Abstract:Recent progress in Neural Causal Models (NCMs) showcased how identification and partial identification of causal effects can be automatically carried out via training of neural generative models that respect the constraints encoded in a given causal graph [Xia et al. 2022, Balazadeh et al. 2022]. However, formal consistency of these methods has only been proven for the case of discrete variables or only for linear causal models. In this work, we prove consistency of partial identification via NCMs in a general setting with both continuous and categorical variables. Further, our results highlight the impact of the design of the underlying neural network architecture in terms of depth and connectivity as well as the importance of applying Lipschitz regularization in the training phase. In particular, we provide a counterexample showing that without Lipschitz regularization the NCM may not be asymptotically consistent. Our results are enabled by new results on the approximability of structural causal models via neural generative models, together with an analysis of the sample complexity of the resulting architectures and how that translates into an error in the constrained optimization problem that defines the partial identification bounds.
Abstract:RLHF has emerged as a pivotal step in aligning language models with human objectives and values. It typically involves learning a reward model from human preference data and then using reinforcement learning to update the generative model accordingly. Conversely, Direct Preference Optimization (DPO) directly optimizes the generative model with preference data, skipping reinforcement learning. However, both RLHF and DPO assume uniform preferences, overlooking the reality of diverse human annotators. This paper presents a new method to align generative models with varied human preferences. We propose an Expectation-Maximization adaptation to DPO, generating a mixture of models based on latent preference types of the annotators. We then introduce a min-max regret ensemble learning model to produce a single generative method to minimize worst-case regret among annotator subgroups with similar latent factors. Our algorithms leverage the simplicity of DPO while accommodating diverse preferences. Experimental results validate the effectiveness of our approach in producing equitable generative policies.