As a vital part of urban public transportation system, subway passenger flow prediction plays a crucial role in alleviating traffic congestion, improving transportation infrastructure, and optimizing travel experience. Existing subway passenger flow prediction mainly focuses on short-term predictions of inbound/outbound passenger flow and origin-destination (O-D) demand. Subway section passenger flow prediction can provide a more direct reflection of passenger fluctuations across different line segments, and offer robust support for management and resource allocation. We propose a subway section passenger flow generation model and a prediction method based on LTiT (LSTM-TSSA-iTransformer). This model is based on the overall architecture of the iTransformer encoder, and an LSTM (Long Short-Term Memory) network is employed to capture the temporal characteristics of subway section passenger flow. This is combined with the TSSA (Token Statistics Self-Attention) to adaptively weight the information at key time points. Efficient performance of the model was evaluated by comparing its predictions with other models, including SARIMA (Seasonal Auto-Regressive integrated moving average), BP neural networks, LightGBM (Light Gradient Boosting Machine) and LSTM (Long Short-Term Memory). Experimental results show that the proposed model outperforms traditional baseline models in evaluation metrics such as R², MAE, MSE, and MAPE. Finally, we further investigate the selection of input window length and prediction step size, and perform robustness analysis under different noise conditions.

Autonomous driving technology heavily depends on accurate and smooth path tracking. Facing complex urban driving scenarios, developing a suite of high-performance and robust parameters for controllers becomes imperative. This paper proposes a stochastic simulation-based optimization model for optimizing the Proportional–Integral–Differential (PID) controller parameters, with tracking accuracy and smoothness as bi-objectives, and solves it using a domination-measure-based efficient global optimization (DMEGO) algorithm. In this model, the tracking accuracy and smoothness are indexed by the normalized dynamic time warping (NDTW) and the mean absolute lateral acceleration (MALA), respectively. In addition, we execute the PID controller in a realistic simulation environment using a CARLA simulator, which consider various city scenes, diverse routes, different vehicle types, road slopes, etc., to provide a comprehensive and reliable evaluation for the designed PID controller. In the DMEGO method, each solution undergoes evaluation using a fixed number of costly simulations. Then, utilizing the solutions and their estimated bi-objective values, two surrogate models for the bi-objectives are constructed using the Gaussian process (GP) model. The preliminary nondominated solutions can be obtained by optimizing the two surrogate models. Finally, a novel performance metric known as the domination measure is employed to evaluate the quality of each solution. This metric is then integrated with the crowding distance to selectively retain a candidate solution exhibiting superior performance and good diversity for the next iteration. In our numerical experiments, we first test the DMEGO algorithm against three other counterparts using a stochastic FON benchmark. The proposed approach is then employed to optimize the PID parameters considering the complexity and uncertainty of urban traffic. The numerical results demonstrate that the nondominated solutions obtained by DMEGO exhibit excellent performance in terms of tracking accuracy and smoothness under limited simulation budgets. Overall, the proposed approach may be a viable tool for solving multi-objective simulation-based optimization problem under uncertainties.

Deep neural networks (DNNs), due to their over-parameterization and memorization capacity, are prone to fitting incorrect labels in training data, which can substantially degrade their performance. Consequently, LNL has attracted increasing attention. Popular LNL methods select potentially clean samples for reliable supervision while applying robust techniques to the remaining samples to mitigate the adverse effects of label noise. However, when sample selection relies solely on the continuously updated classifier, the classifier may gradually memorize incorrect labels during training, and the resulting biased predictions can further affect subsequent sample selection, leading to a self-reinforcing confirmation bias. To address this issue, we propose CAMN, a CLIP-guided adaptive sample selection with machine unlearning and negative learning strategies for noisy labels learning. Specifically, CAMN introduces a pretrained vision language model CLIP as an external semantic prior independent of the task-specific classifier to perform preliminary noise-aware sample selection. The resulting selections are then integrated with an adaptive and balanced sample selection strategy to provide more reliable samples for robust learning. CLIP provides an independent external semantic prior for reducing classifier-induced bias, while the selection results from the trained DNN complements CLIP-based zero-shot selection by reducing over-reliance on its predictions and promoting more balanced sample selection. After selection, CAMN further incorporates a machine unlearning mechanism and a negative learning strategy as regularizers, reducing the persistent influence of noisy labels on DNN and further enhancing its robustness. Experiments demonstrate the effectiveness and robustness of CAMN across diverse noise settings, including synthetic and Web-crawled noisy datasets.