Causal reasoning in natural language requires identifying relevant variables, understanding their interactions, and reasoning about effects and interventions, often under noisy or ambiguous conditions.While large language models (LLMs) exhibit strong general reasoning abilities, they struggle to disentangle correlation from causation, particularly when observations are partially incorrect or irrelevant information is present.In this work, we introduce Noisy-Causal, a new benchmark designed to evaluate causal reasoning under structured noise.Each instance is generated from a ground-truth causal graph and contextualized with a natural language scenario by injecting controllable forms of noise, such as irrelevant distractors, value perturbations, confounding, and partial observability.Moreover, we propose a modular reasoning framework that combines LLMs with explicit causal structure to address these challenges.Our method prompts the LLM to extract variables, construct a causal graph from context, and then reformulates the reasoning task as a structured prompt grounded in this graph.Rather than relying on statistical patterns alone, the LLM is guided by symbolic structure, enabling more interpretable and robust inference.Experimental results show that our method significantly outperforms standard prompting and reasoning baselines on Noisy-Causal.Furthermore, it generalizes well to external benchmarks such as Cladder without task-specific tuning.Our findings highlight the importance of combining causal abstractions with language-driven reasoning to achieve faithful and robust causal understanding in LLMs.

We focus on controllable disentangled representation learning (C-Dis-RL), where users can control the partition of the disentangled latent space to factorize dataset attributes (concepts) for downstream tasks. Two general problems remain under-explored in current methods: (1) They lack comprehensive disentanglement constraints, especially missing the minimization of mutual information between different attributes across latent and observation domains. (2) They lack convexity constraints, which is important for meaningfully manipulating specific attributes for downstream tasks. To encourage both comprehensive C-Dis-RL and convexity simultaneously, we propose a simple yet efficient method: Controllable Interpolation Regularization (CIR), which creates a positive loop where disentanglement and convexity can help each other. Specifically, we conduct controlled interpolation in latent space during training, and we reuse the encoder to help form a ’perfect disentanglement’ regularization. In that case, (a) disentanglement loss implicitly enlarges the potential understandable distribution to encourage convexity; (b) convexity can in turn improve robust and precise disentanglement. CIR is a general module and we merge CIR with three different algorithms: ELEGANT, I2I-Dis, and GZS-Net to show the compatibility and effectiveness. Qualitative and quantitative experiments show improvement in C-Dis-RL and latent convexity by CIR. This further improves downstream tasks: controllable image synthesis, cross-modality image translation and zero-shot synthesis.

Automatic emotion recognition methods are sensitive to the variations across humans and datasets and their performance drops when evaluated across corpora. Domain adaptation (DA) techniques such as Domain-Adversarial Neural Network (DANN) can mitigate this problem. However, domain adaptation cannot guarantee to preserve local features necessary for emotion recognition while reducing domain discrepancies in global features. In this paper, we propose Face wArping emoTion rEcognition (FATE) to address this problem. Unlike the traditional DA models in which the base model is first trained with the source data and then fine-tuned with the source and target data, we reverse the training order. Specifically, we employ first-order facial animation warping to generate a synthetic dataset and utilize contrastive learning to pre-train the encoder. Then, we fine-tune the encoder and the classifier with the source data. After fine-tuning, the model achieves superior emotion recognition performance by preserving the subtle facial features. Our experiments on cross-domain emotion recognition with facial behaviors (Aff-Wild2, SEWA, and SEMAINE) indicate that the proposed FATE model substantially outperforms the domain adaptation models, suggesting that FATE has a better domain generalizability for emotion recognition.

Despite substantial progress in applying neural networks (NN) to a wide variety of areas, they still largely suffer from a lack of transparency and interpretability. While recent developments in explainable artificial intelligence attempt to bridge this gap (e.g., by visualizing the correlation between input pixels and final outputs), these approaches are limited to explaining low-level relationships, and crucially, do not provide insights on error correction. In this work, we propose a framework (VRX) to interpret classification NNs with intuitive structural visual concepts. Given a trained classification model, the proposed VRX extracts relevant class-specific visual concepts and organizes them using structural concept graphs (SCG) based on pairwise concept relationships. By means of knowledge distillation, we show VRX can take a step towards mimicking the reasoning process of NNs and provide logical, concept-level explanations for final model decisions. With extensive experiments, we empirically show VRX can meaningfully answer "why" and "why not" questions about the prediction, providing easy-to-understand insights about the reasoning process. We also show that these insights can potentially provide guidance on improving NN’s performance.