Large language models (LLMs) have shown impressive capabilities in natural language tasks, yet they continue to struggle with multi-step mathematical reasoning, where correctness depends on a precise chain of intermediate steps. Preference optimization methods such as Direct Preference Optimization (DPO) have improved answer-level alignment, but they often overlook the reasoning process itself, providing little supervision over intermediate steps that are critical for complex problem-solving. Existing fine-grained approaches typically rely on strong annotators or reward models to assess the quality of individual steps. However, reward models are vulnerable to reward hacking. To address this, we propose ISLA, a reward-model-free framework that constructs step-level preference data directly from SFT gold traces. ISLA also introduces a self-improving pruning mechanism that identifies informative steps based on two signals: their marginal contribution to final accuracy (relative accuracy) and the model’s uncertainty, inspired by the concept of information gain. Empirically, ISLA achieves better performance than DPO while using only 12% of the training tokens, demonstrating that careful step-level selection can significantly improve both reasoning accuracy and training efficiency.

Zexu Sun, Yongcheng Zeng, Erxue Min, Heyang Gao, Bokai Ji, Dugang Liu, Xing Tang, Xiuqiang He, Xu Chen. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Reinforcement Learning from Human Feedback (RLHF) has emerged as a standard and effective approach for training large language models (LLMs) with human preferences. In this framework, a learned reward model approximates human preferences and guides policy optimization, making it crucial to develop an accurate reward model. However, without the ``true'' reward function, challenges arise when the reward model is an imperfect proxy for human preference. Since the policy optimization continuously shifts the human preference training dataset's distribution. The fixed reward model suffers from this problem of off-distribution, especially the on policy methods. While collecting new preference data can mitigate this issue, it is costly and challenging to optimize. Thus, reusing the policy interaction samples becomes a possible way to further refine the reward model. To tackle these challenges, we introduce a novel method \textbf{U}ncertainty-\textbf{G}radient based \textbf{D}ata \textbf{A}ugmentation (\textbf{UGDA} for short) to enhance reward modeling by leveraging policy samples to maintain on-distribution performance. Specifically, UGDA selects interaction samples based on the uncertainty of the reward ensembles and the gradient based influence of policy optimization. After the reward relabeling of selected samples, we use supervised learning to refine the reward ensembles, then get the retrained policy. Extensive experiments demonstrate that by leveraging UGDA to select a few samples without the costly human preference data collection, we can improve the ability of the policy and surpass the state-of-the-art methods.

In online platforms, incentives (e.g., discounts, coupons) are used to boost user engagement and revenue. Uplift modeling methods are developed to estimate user responses from observational data, often incorporating distribution balancing to address selection bias. However, these methods are limited by in-distribution testing data, which mirrors the training data distribution. In reality, user features change continuously due to time, geography, and other factors, especially on complex online marketing platforms. Thus, effective uplift modeling method for out-of-distribution data is crucial. To address this, we propose a novel uplift modeling method Invariant Deep Uplift Modeling, namely IDUM, which uses invariant learning to enhance out-of-distribution generalization by identifying causal factors that remain consistent across domains. IDUM further refines these features into necessary and sufficient factors and employs a masking component to reduce computational costs by selecting the most informative invariant features. A balancing discrepancy component is also introduced to mitigate selection bias in observational data. We conduct extensive experiments on public and real-world datasets to demonstrate IDUM’s effectiveness in both in-distribution and out-of-distribution scenarios in online marketing. Furthermore, we also provide theoretical analysis and related proofs to support our IDUM’s generalizability.

Uplift modeling is crucial for identifying individuals likely to respond to a treatment in applications like marketing and customer retention, but evaluating these models is challenging due to the inaccessibility of counterfactual outcomes in real-world settings. In this paper, we identify a fundamental limitation in existing evaluation metrics, such as the uplift and Qini curves, which fail to rank individuals with binary negative outcomes accurately. This can lead to biased evaluations, where biased models receive higher curve values than unbiased ones, resulting in suboptimal model selection. To address this, we propose the Principled Uplift Curve (PUC), a novel evaluation metric that assigns equal curve values of individuals with both positive and negative binary outcomes, offering a more balanced and unbiased assessment. We then derive the Principled Uplift Loss (PUL) function from the PUC and integrate it into a new uplift model, the Principled Treatment and Outcome Network (PTONet), to reduce bias during uplift model training. Experiments on both simulated and real-world datasets demonstrate that the PUC provides less biased evaluations, while PTONet outperforms existing methods. The source code is available at: https://github.com/euzmin/PUC.

Improving user engagement and platform revenue is crucial for online marketing platforms. Uplift modeling is proposed to solve this problem, which applies different treatments (e.g., discounts, bonus) to satisfy corresponding users. Despite progress in this field, limitations persist. Firstly, most of them focus on scenarios where only user features exist. However, in real-world scenarios, there are rich contexts available in the online platform (e.g., short videos, news), and the uplift model needs to infer an incentive for each user on the specific item, which is called real-time marketing. Thus, only considering the user features will lead to biased prediction of the responses, which may cause the cumulative error for uplift prediction. Moreover, due to the large-scale contexts, directly concatenating the context features with the user features will cause a severe distribution shift in the treatment and control groups. Secondly, capturing the interaction relationship between the user features and context features can better predict the user response. To solve the above limitations, we propose a novel model-agnostic Robust Uplift Modeling with Large-Scale Contexts (UMLC) framework for Real-time Marketing. Our UMLC includes two customized modules. 1) A response-guided context grouping module for extracting context features information and condensing value space through clusters. 2) A feature interaction module for obtaining better uplift prediction. Specifically, this module contains two parts: a user-context interaction component for better modeling the response; a treatment-feature interaction component for discovering the treatment assignment sensitive feature of each instance to better predict the uplift. Moreover, we conduct extensive experiments on a synthetic dataset and a real-world product dataset to verify the effectiveness and compatibility of our UMLC.

Large language models (LLMs) have demonstrated significant improvements in contextual understanding. However, their ability to attend to truly critical information during long-context reasoning and generation still falls behind the pace. Specifically, our preliminary experiments reveal that certain distracting patterns can misdirect the model’s attention during inference, and removing these patterns substantially improves reasoning accuracy and generation quality. We attribute this phenomenon to spurious correlations in the training data, which obstruct the model’s capacity to infer authentic causal instruction–response relationships. This phenomenon may induce redundant reasoning processes, potentially resulting in significant inference overhead and, more critically, the generation of erroneous or suboptimal responses. To mitigate this, we introduce a two-stage framework called Learning to Focus (LeaF) leveraging intervention-based inference to disentangle confounding factors. In the first stage, LeaF employs gradient-based comparisons with an advanced teacher to automatically identify confounding tokens based on causal relationships in the training corpus. Then, in the second stage, it prunes these tokens during distillation to enact intervention, aligning the student’s attention with the teacher’s focus distribution on truly critical context tokens. Experimental results demonstrate that LeaF not only achieves an absolute improvement in various mathematical reasoning, code generation and multi-hop question answering benchmarks but also effectively suppresses attention to confounding tokens during inference, yielding a more interpretable and reliable reasoning model.

Sequential causal effect estimation has recently attracted increasing attention from research and industry. While the existing models have achieved many successes, there are still many limitations. Existing models usually assume the causal graphs to be sufficient, i.e., there are no latent factors, such as the unmeasured confounders and instrumental variables. However, in real-world scenarios, it is hard to record all of the factors in the observational data, which makes the causally sufficient assumptions not hold. Moreover, existing models mainly focus on discrete treatments rather than continuous ones. To alleviate the above problems, in this paper, we propose a novelContinousCausalModel by explicitly capturing theLatentFactors (calledC$^{2}$2M-LFfor short). Specifically, we define a sequential causal graph by simultaneously considering the unmeasured confounders and instrumental variables. Second, we describe the independence that should be satisfied among different variables from the mutual information perspective and further propose our learning objective. Then, we reweight different samples in the continuous treatment space to optimize our model unbiasedly. Beyond the above designs, we also theoretically analyze our model’s causal identifiability and unbiasedness. Finally, we conduct extensive experiments on both simulation and real-world datasets to demonstrate the effectiveness of our proposed model.

Customer Lifetime Value (CLTV) prediction is a critical task in business applications, such as customer relationship management (CRM), online marketing, etc. Accurately predicting CLTV is challenging in real-world business scenarios, as the distribution of CLTV is complex and mutable. Firstly, there is a large number of users without any consumption consisting of a long-tailed part that is too complex to fit. Secondly, the small set of high-value users spent orders of magnitude more than a typical user leading to a wide range of the CLTV distribution which is hard to capture in a single distribution. Existing approaches for CLTV estimation either assume a prior probability distribution and fit a single group of distribution-related parameters for all samples, or directly learn from the posterior distribution with manually predefined buckets in a heuristic manner. However, all these methods fail to handle complex and mutable distributions. In this paper, we propose a novel optimal distribution selection model (OptDist) for CLTV prediction, which utilizes an adaptive optimal sub-distribution selection mechanism to improve the accuracy of complex distribution modeling. Specifically, OptDist trains several candidate sub-distribution networks in the distribution learning module (DLM) for modeling the probability distribution of CLTV. Then, a distribution selection module (DSM) is proposed to select the sub-distribution for each sample, thus making the selection automatically and adaptively. Besides, we design an alignment mechanism that connects both modules, which effectively guides the optimization. We conduct extensive experiments on both two public and one private dataset to verify that OptDist outperforms state-of-the-art baselines. Furthermore, OptDist has been deployed on a large-scale financial platform for customer acquisition marketing campaigns and the online experiments also demonstrate the effectiveness of OptDist.

Yiju Guo, Ganqu Cui, Lifan Yuan, Ning Ding, Zexu Sun, Bowen Sun, Huimin Chen, Ruobing Xie, Jie Zhou, Yankai Lin, Zhiyuan Liu, Maosong Sun. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

Uplift modeling is a technique used to predict the effect of a treatment (e.g., discounts) on an individual’s response. Although several methods have been proposed for multi-valued treatment, they are extended from binary treatment methods. There are still some limitations. Firstly, existing methods calculate uplift based on predicted responses, which may not guarantee a consistent uplift distribution between treatment and control groups. Moreover, this may cause cumulative errors for multi-valued treatment. Secondly, the model parameters become numerous with many prediction heads, leading to reduced efficiency. To address these issues, we propose a novel Multi-gate Mixture-of-Experts based Multi-valued Treatment Network (M3TN). M3TN consists of two components: 1) a feature representation module with Multi-gate Mixture-of-Experts to improve the efficiency; 2) a reparameterization module by modeling uplift explicitly to improve the effectiveness. We also conduct extensive experiments to demonstrate the effectiveness and efficiency of our M3TN.

Causal discovery with observational and interventional data plays an important role in numerous fields. Due to the costly and potentially risky nature of intervention experiments, selecting informative interventions is critical in real-world situations. Several recent works introduce Bayesian active learning to select interventions that maximize the expected information gain about the underlying causal relationship at each optimization step. However, there are still some limitations within these methods: (1) Local optimality. With multiple intervention experiments, selecting optimal intervention myopically at each step may drop into the local optimal point. (2) Expensive time cost. Optimizing the most informative intervention at each step is time-consuming and not suitable for adaptive experiments with strict inference speed requirements. In this study, we propose a novel method called Reinforcement Learning-based Causal Bayesian Experimental Design (RL-CBED) to reduce the risk of local optimality and accelerate intervention selection inference. Specifically, we formulate the active causal discovery problem as a partially observable Markov decision process (POMDP). We design an information gain-based sparse reward function and then improve it to a dense reward function, providing fine-grained feedback to help the RL policy learn more quickly in complex environments. Moreover, we theoretically prove that the Q-function estimator can be learned using only trajectories sampled from the prior, which can significantly reduce the time cost of training process, enabling the real-world application of our method. Extensive experiments on both synthetic and real world-inspired semi-synthetic datasets demonstrate the effectiveness of our proposed method.

Uplift modeling has been widely employed in online marketing by predicting the response difference between the treatment and control groups, so as to identify the sensitive individuals toward interventions like coupons or discounts. Compared with traditional conversion uplift modeling,revenue uplift modeling exhibits higher potential due to its direct connection with the corporate income. However, previous works can hardly handle the continuous long-tail response distribution in revenue uplift modeling. Moreover, they have neglected to optimize the uplift ranking among different individuals, which is actually the core of uplift modeling. To address such issues, in this paper, we first utilize the zero-inflated lognormal (ZILN) loss to regress the responses and customize the corresponding modeling network, which can be adapted to different existing uplift models. Then, we study the ranking-related uplift modeling error from the theoretical perspective and propose two tighter error bounds as the additional loss terms to the conventional response regression loss. Finally, we directly model the uplift ranking error for the entire population with a listwise uplift ranking loss. The experiment results on offline public and industrial datasets validate the effectiveness of our method for revenue uplift modeling. Furthermore, we conduct large-scale experiments on a prominent online fintech marketing platform, Tencent FiT, which further demonstrates the superiority of our method in real-world applications.

In recent years, graph contrastive learning (GCL) has received increasing attention in recommender systems due to its effectiveness in reducing bias caused by data sparsity. However, most existing GCL models rely on heuristic approaches and usually assume entity independence when constructing contrastive views. We argue that these methods struggle to strike a balance between semantic invariance and view hardness across the dynamic training process, both of which are critical factors in graph contrastive learning. To address the above issues, we propose a novel GCL-based recommendation framework RGCL, which effectively maintains the semantic invariance of contrastive pairs and dynamically adapts as the model capability evolves through the training process. Specifically, RGCL first introduces decision boundary-aware adversarial perturbations to constrain the exploration space of contrastive augmented views, avoiding the decrease of task-specific information. Furthermore, to incorporate global user-user and item-item collaboration relationships for guiding on the generation of hard contrastive views, we propose an adversarial-contrastive learning objective to construct a relation-aware view-generator. Besides, considering that unsupervised GCL could potentially narrower margins between data points and the decision boundary, resulting in decreased model robustness, we introduce the adversarial examples based on maximum perturbations to achieve margin maximization. We also provide theoretical analyses on the effectiveness of our designs. Through extensive experiments on five public datasets, we demonstrate the superiority of RGCL compared against twelve baseline models.

In modern online platforms, incentives (e.g., discounts, bonus) are essential factors that enhance user engagement and increase platform revenue. Over recent years, uplift modeling has been introduced as a strategic approach to assign incentives to individual customers. Especially in many real-world applications, online platforms can only incentivize customers with specific budget constraints. This problem can be reformulated as the multi-choice knapsack problem (MCKP). The objective of this optimization is to select the optimal incentive for each customer to maximize the return on investment (ROI). Recent works in this field frequently tackle the budget allocation problem using a two-stage approach. However, this solution is confronted with the following challenges: (1) The causal inference methods often ignore the domain knowledge in online marketing, where the expected response curve of a customer should be monotonic and smooth as the incentive increases. (2) There is an optimality gap between the two stages, resulting in inferior sub-optimal allocation performance due to the loss of the incentive recommendation information for the uplift prediction under the limited budget constraint. To address these challenges, we propose a novel End-to-End Cost-Effective Incentive Recommendation (E3IR) model under the budget constraint. Specifically, our methods consist of two modules, i.e., the uplift prediction module and the differentiable allocation module. In the uplift prediction module, we construct prediction heads to capture the incremental improvement between adjacent treatments with the marketing domain constraints (i.e., monotonic and smooth). We incorporate integer linear programming (ILP) as a differentiable layer input in the differentiable allocation module. Furthermore, we conduct extensive experiments on public and real product datasets, demonstrating that our E3IR improves allocation performance compared to existing two-stage approaches.

Uplift modeling has shown very promising results in online marketing. However, most existing works are prone to the robustness challenge in some practical applications. In this paper, we first present a possible explanation for the above phenomenon. We verify that there is a feature sensitivity problem in online marketing using different real-world datasets, where the perturbation of some key features will seriously affect the performance of the uplift model and even cause the opposite trend. To solve the above problem, we propose a novel robustness-enhanced uplift modeling framework with adversarial feature desensitization (RUAD). Specifically, our RUAD can more effectively alleviate the feature sensitivity of the uplift model through two customized modules, including a feature selection module with joint multi-label modeling to identify a key subset from the input features and an adversarial feature desensitization module using adversarial training and soft interpolation operations to enhance the robustness of the model against this selected subset of features. Finally, we conduct extensive experiments on a public dataset and a real product dataset to verify the effectiveness of our RUAD in online marketing. In addition, we also demonstrate the robustness of our RUAD to the feature sensitivity, as well as the compatibility with different uplift models.

In offline imitation learning (IL), an agent aims to learn an optimal expert behavior policy without additional online environment interactions. However, in many real-world scenarios, such as robotics manipulation, the offline dataset is collected from suboptimal behaviors without rewards. Due to the scarce expert data, the agents usually suffer from simply memorizing poor trajectories and are vulnerable to the variations in the environments, lacking the capability of generalizing to new environments.To automatically generate high-quality expert data and improve the generalization ability of the agent, we propose a framework named \underline{O}ffline \underline{I}mitation \underline{L}earning with \underline{C}ounterfactual data \underline{A}ugmentation (OILCA) by doing counterfactual inference. In particular, we leverage identifiable variational autoencoder to generate \textit{counterfactual} samples for expert data augmentation. We theoretically analyze the influence of the generated expert data and the improvement of generalization. Moreover, we conduct extensive experiments to demonstrate that our approach significantly outperforms various baselines on both \textsc{DeepMind Control Suite} benchmark for in-distribution performance and \textsc{CausalWorld} benchmark for out-of-distribution generalization.