Recent advances in large language models (LLMs) have broadened their applicability across diverse tasks, yet specialized domains still require targeted post-training. Among existing methods, Group Relative Policy Optimization (GRPO) stands out for its efficiency, leveraging groupwise relative rewards while avoiding costly value function learning. However, GRPO treats candidate responses as independent, overlooking semantic interactions such as complementarity and contradiction. To address this challenge, we first introduce a Structural Causal Model (SCM) that reveals hidden dependencies among candidate responses induced by conditioning on a final integrated output, forming a collider structure. Then, our causal analysis leads to two insights: (1) projecting responses onto a causally-informed subspace improves prediction quality, and (2) this projection yields a better baseline than query-only conditioning. Building on these insights, we propose Group Causal Policy Optimization (GCPO), which integrates causal structure into optimization through two key components: a causally-informed reward adjustment and a novel KL-regularization term that aligns the policy with a causally-projected reference distribution. Comprehensive experimental evaluations on various benchmarks demonstrate that GCPO consistently surpasses existing methods.

External reasoning systems combine language models with process reward models (PRMs) to select high-quality reasoning paths for complex tasks such as mathematical problem solving. However, these systems are prone to reward hacking, where high-scoring but logically incorrect paths are assigned high scores by the PRMs, leading to incorrect answers. From a causal inference perspective, we attribute this phenomenon primarily to the presence of confounding semantic features. To address it, we propose Causal Reward Adjustment (CRA), a method that mitigates reward hacking by estimating the true reward of a reasoning path. CRA trains sparse autoencoders on the PRM’s internal activations to recover interpretable features, then corrects confounding by using backdoor adjustment. Experiments on math solving datasets demonstrate that CRA mitigates reward hacking and improves final accuracy, without modifying the policy model or retraining PRM.

In this paper, we focus on the out-of-distribution (OOD) generalization of self-supervised learning (SSL). By analyzing the mini-batch construction during the SSL training phase, we first give one plausible explanation for SSL having OOD generalization. Then, from the perspective of data generation and causal inference, we analyze and conclude that SSL learns spurious correlations during the training process, which leads to a reduction in OOD generalization. To address this issue, we propose a post-intervention distribution (PID) grounded in the Structural Causal Model. PID offers a scenario where the spurious variable and label variable is mutually independent. Besides, we demonstrate that if each mini-batch during SSL training satisfies PID, the resulting SSL model can achieve optimal worst-case OOD performance. This motivates us to develop a batch sampling strategy that enforces PID constraints through the learning of a latent variable model. Through theoretical analysis, we demonstrate the identifiability of the latent variable model and validate the effectiveness of the proposed sampling strategy. Experiments conducted on various downstream OOD tasks demonstrate the effectiveness of the proposed sampling strategy.

Contrastive Language-Image Pretraining (CLIP) has achieved remarkable success, but its performance can degrade when fine-tuned in out-of-distribution (OOD) scenarios. We model the prediction process using a Structural Causal Model (SCM) and show that the causal mechanism involving both invariant and variant factors in training environments differs from that in test environments. In contrast, the causal mechanism with solely invariant factors remains consistent across environments. We theoretically prove the existence of a linear mapping from CLIP embeddings to invariant factors, which can be estimated using interventional data. Additionally, we provide a condition to guarantee low OOD risk of the invariant predictor. Based on these insights, we propose the Invariant Causal Mechanism of CLIP (CLIP-ICM) framework. CLIP-ICM involves collecting interventional data, estimating a linear projection matrix, and making predictions within the invariant subspace. Experiments on several OOD datasets show that CLIP-ICM significantly improves the performance of CLIP. Our method offers a simple but powerful enhancement, boosting the reliability of CLIP in real-world applications.

Large language models (LLMs) excel at complex tasks thanks to advances in their reasoning abilities. However, existing methods overlook the trade-off between reasoning effectiveness and efficiency, often encouraging unnecessarily long reasoning chains and wasting tokens. To address this, we propose Learning to Think (L2T), an information-theoretic reinforcement fine-tuning framework for LLMs to make the models achieve optimal reasoning with fewer tokens. Specifically, L2T treats each query-response interaction as a hierarchical session of multiple episodes and proposes a universal dense process reward, i.e., quantifies the episode-wise information gain in parameters, requiring no extra annotations or task-specific evaluators. We propose a method to quickly estimate this reward based on PAC-Bayes bounds and the Fisher information matrix. Theoretical analyses show that it significantly reduces computational complexity with high estimation accuracy. By immediately rewarding each episode's contribution and penalizing excessive updates, L2T optimizes the model via reinforcement learning to maximize the use of each episode and achieve effective updates. Empirical results on various reasoning benchmarks and base models demonstrate the advantage of L2T across different tasks, boosting both reasoning effectiveness and efficiency.

Long-term time series forecasting is a long-standing challenge in various applications. A central issue in time series forecasting is that methods should expressively capture long-term dependency. Furthermore, time series forecasting methods should be flexible when applied to different scenarios. Although Fourier analysis offers an alternative to effectively capture reusable and periodic patterns to achieve long-term forecasting in different scenarios, existing methods often assume high-frequency components represent noise and should be discarded in time series forecasting. However, we conduct a series of motivation experiments and discover that the role of certain frequencies varies depending on the scenarios. In some scenarios, removing high-frequency components from the original time series can improve the forecasting performance, while in others scenarios, removing them is harmful to forecasting performance. Therefore, it is necessary to treat the frequencies differently according to specific scenarios. To achieve this, we first reformulate the time series forecasting problem as learning a transfer function of each frequency in the Fourier domain. Further, we design Frequency Dynamic Fusion (FreDF), which individually predicts each Fourier component, and dynamically fuses the output of different frequencies. Moreover, we provide a novel insight into the generalization ability of time series forecasting and propose the generalization bound of time series forecasting. Then we prove FreDF has a lower bound, indicating that FreDF has better generalization ability. Extensive experiments conducted on multiple benchmark datasets and ablation studies demonstrate the effectiveness of FreDF.

Remote sensing image (RSI) classification is a popular research topic that aims to assign semantic labels to images acquired from aerial or maritime platforms. Existing deep feature learning methods for this task can be divided into two paradigms: generative and discriminative. The former methods are good at capturing every local detail of images, while the later approaches focus on the most salient area. The significant differences between the two types of methods, both in terms of their underlying mechanisms and practical implementation, motivate us to integrate information acquired by both paradigms by exploiting their complementary strengths. However, this idea faces a challenge that local information in the extracted features, especially those from generative methods, may not be reliable for RSI classification. The reason for this challenge is that, due to the characteristics of the ground observation perspective, some RSIs, while semantically different, exhibit a significant degree of similarity in local details. This phenomenon leads to insufficient discriminability of local features to separate multiple RSI categories, which implies that the classification results overly focused on local information may be unreliable. To address this issue, in this article, we propose a novel framework that integrates generative and discriminative feature learning methods with evidential learning for RSI classification. Our framework uses the Dirichlet distribution to model the predicted probabilities to be integrated, thereby collecting evidence about their reliability. This enables us to integrate multiple features at an evidence level and make reliable decisions, overcoming the unreliabilities of generative-discriminative joint feature learning induced by RSI characteristics. We evaluate the proposed framework on several satellite and shipborne RSI classification datasets. The experimental results show that our method outperforms the state-of-the-art baselines in terms of accuracy and robustness.