This paper aims to effectively reduce CO2 emissions by examining the impact of three distinct incentive and constraint policies on the quality of rating and certification information in China’s green bond issuance market. To accomplish this, the government has implemented incentives, while regulators have introduced constraints to curb the spread of inflated rating and certification information. We build on the integrated rating and certification regulation mechanism by presenting a two-stage Stackelberg game model that involves four key participants: the China Securities Regulatory Commission, local governments, green evaluation and certification agencies, and credit rating agencies. We incorporate environmental effects indicators into the expected utility of rating and certification agencies to investigate the equilibrium conditions under three policy scenarios: a single financial incentive policy, a single regulatory constraint policy, and a combined incentive and constraint policy. The paper employs Stackelberg game theory to analyze how different policies mitigate the occurrence of “inflated” ratings and “greenwashing” in certifications. Numerical analysis is conducted to validate the theoretical findings. Moreover, we assess the impact of these policies on the quality of rating and evaluation information, using data from China’s green bond issuance market between 2016 and 2021. Our research offers valuable management insights and regulatory recommendations for both regulators and local governments.

The rapid development of social media has driven the need for opinion mining and sentiment analysis based on multimodal samples. As a fine-grained task within multimodal sentiment analysis, aspect-based multimodal sentiment analysis (ABMSA) enables the accurate and efficient determination of sentiment polarity for aspect-level targets. However, traditional ABMSA methods often perform suboptimally on social media samples, as the images in these samples typically contain embedded text that conventional models overlook. Such text influences sentiment judgment. To address this issue, we propose a text-in-image enhanced self-supervised alignment model (TESAM) that accounts for multimodal information more comprehensively. Specifically, we employed Optical Character Recognition technology to extract embedded text from images and, based on the principle that text-in-image is an integral part of the visual modality, fused it with visual features to obtain more comprehensive image representations. Additionally, we incorporate aspect words to guide the model in disregarding irrelevant semantic features, thereby reducing noise interference. Furthermore, to mitigate the semantic gap between modalities, we propose pre-training the feature extraction module with self-supervised alignment. During this pre-training stage, unimodal semantic embeddings from both modalities are aligned by calculating errors using Euclidean distance and cosine similarity. Experimental results demonstrate that TESAM achieved remarkable performances on three ABMSA benchmarks. These results validate the rationale and effectiveness of our proposed improvements.

In-the-wild facial expression recognition (FER) is inherently affected by ambiguous expressions and subjective annotations, making the observed labels only partially reliable. Conventional one-hot supervision nevertheless treats these labels as ground truth and may drive deep networks to memorize corrupted annotations. This paper proposes RDAS, a reliability-driven adaptive supervision framework for noisy-label FER. The central idea was to use label reliability not merely as a sample weight, but as a variable that determines the supervision policy of each training sample. RDAS enhances expression-sensitive representations by combining RGB appearance with structural high-frequency cues, estimates label reliability with an exponential moving average (EMA) teacher through the margin between the annotated class and its strongest competing class, and assigns reliability-specific objectives to high-, medium-, and low-reliability samples. Reliable samples retained hard-label learning, ambiguous samples received teacher-guided soft correction, and unreliable samples were prevented from dominating training with misleading gradients. Experiments on RAF-DB, AffectNet, and FERPlus datasets showed that RDAS improved both standard recognition performance and robustness under synthetic label noise. On RAF-DB, RDAS achieved an average accuracy of $ 89.70\% $ on the original dataset and $ 86.15\% $ under 30% label noise. Ablation studies provided quantitative support for reliability modeling and adaptive supervision, while the visual analyses provided complementary qualitative evidence consistent with the proposed mechanism.