Counterfactual Data Augmentation with Contrastive Learning

Statistical disparity between distinct treatment groups is one of the most significant challenges for estimating Conditional Average Treatment Effects (CATE). To address this, we introduce a model-agnostic data augmentation method that imputes the counterfactual outcomes for a selected subset of individuals. Specifically, we utilize contrastive learning to learn a representation space and a similarity measure such that in the learned representation space close individuals identified by the learned similarity measure have similar potential outcomes. This property ensures reliable imputation of counterfactual outcomes for the individuals with close neighbors from the alternative treatment group. By augmenting the original dataset with these reliable imputations, we can effectively reduce the discrepancy between different treatment groups, while inducing minimal imputation error. The augmented dataset is subsequently employed to train CATE estimation models. Theoretical analysis and experimental studies on synthetic and semi-synthetic benchmarks demonstrate that our method achieves significant improvements in both performance and robustness to overfitting across state-of-the-art models.

Off-Policy Evaluation for Human Feedback

Off-policy evaluation (OPE) is important for closing the gap between offline training and evaluation of reinforcement learning (RL), by estimating performance and/or rank of target (evaluation) policies using offline trajectories only. It can improve the safety and efficiency of data collection and policy testing procedures in situations where online deployments are expensive, such as healthcare. However, existing OPE methods fall short in estimating human feedback (HF) signals, as HF may be conditioned over multiple underlying factors and is only sparsely available; as opposed to the agent-defined environmental rewards (used in policy optimization), which are usually determined over parametric functions or distributions. Consequently, the nature of HF signals makes extrapolating accurate OPE estimations to be challenging. To resolve this, we introduce an OPE for HF (OPEHF) framework that revives existing OPE methods in order to accurately evaluate the HF signals. Specifically, we develop an immediate human reward (IHR) reconstruction approach, regularized by environmental knowledge distilled in a latent space that captures the underlying dynamics of state transitions as well as issuing HF signals. Our approach has been tested over two real-world experiments, adaptive in-vivo neurostimulation and intelligent tutoring, as well as in a simulation environment (visual Q&A). Results show that our approach significantly improves the performance toward estimating HF signals accurately, compared to directly applying (variants of) existing OPE methods.

Robust Reinforcement Learning through Efficient Adversarial Herding

Although reinforcement learning (RL) is considered the gold standard for policy design, it may not always provide a robust solution in various scenarios. This can result in severe performance degradation when the environment is exposed to potential disturbances. Adversarial training using a two-player max-min game has been proven effective in enhancing the robustness of RL agents. In this work, we extend the two-player game by introducing an adversarial herd, which involves a group of adversaries, in order to address ($\textit{i}$) the difficulty of the inner optimization problem, and ($\textit{ii}$) the potential over pessimism caused by the selection of a candidate adversary set that may include unlikely scenarios. We first prove that adversarial herds can efficiently approximate the inner optimization problem. Then we address the second issue by replacing the worst-case performance in the inner optimization with the average performance over the worst-$k$ adversaries. We evaluate the proposed method on multiple MuJoCo environments. Experimental results demonstrate that our approach consistently generates more robust policies.

Few-Shot Continual Learning for Conditional Generative Adversarial Networks

In few-shot continual learning for generative models, a target mode must be learned with limited samples without adversely affecting the previously learned modes. In this paper, we propose a new continual learning approach for conditional generative adversarial networks (cGAN) based on a new mode-affinity measure for generative modeling. Our measure is entirely based on the cGAN's discriminator and can identify the existing modes that are most similar to the target. Subsequently, we expand the continual learning model by including the target mode using a weighted label derived from those of the closest modes. To prevent catastrophic forgetting, we first generate labeled data samples using the cGAN's generator, and then train the cGAN model for the target mode while memory replaying with the generated data. Our experimental results demonstrate the efficacy of our approach in improving the generation performance over the baselines and the state-of-the-art approaches for various standard datasets while utilizing fewer training samples.

PASTA: Pessimistic Assortment Optimization

We consider a class of assortment optimization problems in an offline data-driven setting. A firm does not know the underlying customer choice model but has access to an offline dataset consisting of the historically offered assortment set, customer choice, and revenue. The objective is to use the offline dataset to find an optimal assortment. Due to the combinatorial nature of assortment optimization, the problem of insufficient data coverage is likely to occur in the offline dataset. Therefore, designing a provably efficient offline learning algorithm becomes a significant challenge. To this end, we propose an algorithm referred to as Pessimistic ASsortment opTimizAtion (PASTA for short) designed based on the principle of pessimism, that can correctly identify the optimal assortment by only requiring the offline data to cover the optimal assortment under general settings. In particular, we establish a regret bound for the offline assortment optimization problem under the celebrated multinomial logit model. We also propose an efficient computational procedure to solve our pessimistic assortment optimization problem. Numerical studies demonstrate the superiority of the proposed method over the existing baseline method.

Domain Adaptation via Rebalanced Sub-domain Alignment

Unsupervised domain adaptation (UDA) is a technique used to transfer knowledge from a labeled source domain to a different but related unlabeled target domain. While many UDA methods have shown success in the past, they often assume that the source and target domains must have identical class label distributions, which can limit their effectiveness in real-world scenarios. To address this limitation, we propose a novel generalization bound that reweights source classification error by aligning source and target sub-domains. We prove that our proposed generalization bound is at least as strong as existing bounds under realistic assumptions, and we empirically show that it is much stronger on real-world data. We then propose an algorithm to minimize this novel generalization bound. We demonstrate by numerical experiments that this approach improves performance in shifted class distribution scenarios compared to state-of-the-art methods.

Causal Knowledge Transfer from Task Affinity

Recent developments in deep representation models through counterfactual balancing have led to a promising framework for estimating Individual Treatment Effects (ITEs) that are essential to causal inference in the Neyman-Rubin potential outcomes framework. While Randomized Control Trials are vital to understanding causal effects, they are sometimes infeasible, costly, or unethical to conduct. Motivated by these potential obstacles to data acquisition, we focus on transferring the causal knowledge acquired in prior experiments to new scenarios for which only limited data is available. To this end, we first observe that the absolute values of ITEs are invariant under the action of the symmetric group on the labels of treatments. Given this invariance, we propose a symmetrized task distance for calculating the similarity of a target scenario with those encountered before. The aforementioned task distance is then used to transfer causal knowledge from the closest of all the available previously learned tasks to the target scenario. We provide upper bounds on the counterfactual loss and ITE error of the target task indicating the transferability of causal knowledge. Empirical studies are provided for various real-world, semi-synthetic, and synthetic datasets demonstrating that the proposed symmetrized task distance is strongly related to the estimation of the counterfactual loss. Numerical results indicate that transferring causal knowledge reduces the amount of required data by up to 95% when compared to training from scratch. These results reveal the promise of our method when applied to important albeit challenging real-world scenarios such as transferring the knowledge of treatment effects (e.g., medicine, social policy, personal training, etc.) studied on a population to other groups absent in the study.

Multi-Agent Adversarial Attacks for Multi-Channel Communications

Recently Reinforcement Learning (RL) has been applied as an anti-adversarial remedy in wireless communication networks. However, studying the RL-based approaches from the adversary's perspective has received little attention. Additionally, RL-based approaches in an anti-adversary or adversarial paradigm mostly consider single-channel communication (either channel selection or single channel power control), while multi-channel communication is more common in practice. In this paper, we propose a multi-agent adversary system (MAAS) for modeling and analyzing adversaries in a wireless communication scenario by careful design of the reward function under realistic communication scenarios. In particular, by modeling the adversaries as learning agents, we show that the proposed MAAS is able to successfully choose the transmitted channel(s) and their respective allocated power(s) without any prior knowledge of the sender strategy. Compared to the single-agent adversary (SAA), multi-agents in MAAS can achieve significant reduction in signal-to-noise ratio (SINR) under the same power constraints and partial observability, while providing improved stability and a more efficient learning process. Moreover, through empirical studies we show that the results in simulation are close to the ones in communication in reality, a conclusion that is pivotal to the validity of performance of agents evaluated in simulations.