Driven by advances in artificial intelligence, deep reinforcement learning (DRL) has made remarkable strides in adaptive traffic signal control (ATSC), empowering improved handling of fluctuating traffic volumes and congestion. However, in most existing studies, trained agents exhibit poor transferability in scenarios with varying vehicle turning ratios, and the switching rules for the signal stage sequence do not align with the actual traffic demands. To address these issues, this paper presents action masking based proximal policy optimization with the dual-ring phase structure (AMPPO-DR), a novel ATSC model based on DRL that can simultaneously optimize the stage sequence and duration. Specifically, we consider the correlation between states and actions and utilize intersection channelization to predict vehicle turning directions. Moreover, we define the action as selecting the next green stage and establish variable-stage-sequence constraint rules on the basis of the dual-ring phase structure. To satisfy the constraints on the stage sequence, we propose the AMPPO algorithm, which dynamically adjusts the policy network outputs to mask invalid stages in real time. The simulation experiments demonstrate that the proposed method can effectively adapt to changing turning flows, enabling flexible and rational stage switching and ultimately increasing traffic efficiency.

Multi-view partial multi-label learning (MVPML) aims to learn a multi-label predictive model from the training examples, each of which is presented by multiple feature vectors while associated with a set of candidate labels where only a subset is correct. Generally, existing techniques work simply by identifying the ground-truth label via aggregating the features from all views to train a final classifier, but ignore the cause of the incorrect labels in the candidate label sets, i.e., the diverse property of the representation from different views leads to the incorrect labels which form the candidate labels alone with the essential supervision. In this paper, a novel MVPML approach is proposed to learn the predictive model and the incorrect-labeling model jointly by incorporating the graph-fusion-based topological structure of the feature space. Specifically, the latent label distribution and the incorrect labels are identified simultaneously in a unified framework under the supervision of candidate labels. In addition, a common topological structure of the feature space from all views is learned via the graph fusion for further capturing the latent label distribution. Experimental results on the real-world datasets clearly validate the effectiveness of the proposed approach for solving multi-view partial multi-label learning problems.