The paper "Robust Emotion Recognition in Context Debiasing" addresses the challenge of context bias interference in context-aware emotion recognition (CAER) systems. Context bias can lead models to rely on spurious correlations between background contexts and emotion labels, causing performance bottlenecks. The authors propose a counterfactual emotion inference (CLEF) framework to mitigate this issue. CLEF formulates a generalized causal graph to decouple the causal relationships among variables in CAER. It introduces a non-invasive context branch to capture the adverse direct effect of context bias. During inference, CLEF eliminates the direct context effect by comparing factual and counterfactual outcomes, resulting in unbiased predictions. The framework is model-agnostic and can be integrated into existing CAER methods. Extensive experiments on large-scale datasets show that CLEF consistently improves the performance of various CAER models, demonstrating its effectiveness and broad applicability.The paper "Robust Emotion Recognition in Context Debiasing" addresses the challenge of context bias interference in context-aware emotion recognition (CAER) systems. Context bias can lead models to rely on spurious correlations between background contexts and emotion labels, causing performance bottlenecks. The authors propose a counterfactual emotion inference (CLEF) framework to mitigate this issue. CLEF formulates a generalized causal graph to decouple the causal relationships among variables in CAER. It introduces a non-invasive context branch to capture the adverse direct effect of context bias. During inference, CLEF eliminates the direct context effect by comparing factual and counterfactual outcomes, resulting in unbiased predictions. The framework is model-agnostic and can be integrated into existing CAER methods. Extensive experiments on large-scale datasets show that CLEF consistently improves the performance of various CAER models, demonstrating its effectiveness and broad applicability.