Alcohol Use Disorder (AUD), encompassing Alcohol Abuse (AA) and Alcohol Dependence (AD), is a chronic, relapsing brain condition that can cause mental and physical issues. Confusion between AA and AD can lead to ineffective or even excessively aggressive treatments, but distinguishing them is difficult due to similar symptoms and physiological indicators. Fortunately, from a psychological perspective, AA is closely linked to poor behavioral control, while AD is associated with affective lability. These psychological mechanisms are differently featured in the activity of patients' brain regions. Inspired by this, we propose a model called Multiscale and Hybrid Spatial Temporal Graph Convolutional Network (MuST-GCN) that enables to extract the features of electroencephalogram (EEG) signals for accurate identification of AA and AD within AUD. It includes two modules: Multiscale Feature Extraction (MSFE) module with GCN concurrently analyzes inter-regional and intra-regional connectivity of brain regions to extract EEG features, exploring functional connectivity differences in AA and AD patients; Hybrid Spatial Temporal Memory (HSTM) module integrates spatial and temporal attention mechanisms to refine the significant features extracted from MSFE, targeting the key brain regions and temporal dynamics. The HSTM module yields a refined feature representation reducing overfitting and improving multiclass classification accuracy. MuST-GCN is evaluated using five-fold cross-validation on two datasets, achieving classification accuracies of 90.74% and 99.99%, respectively, and demonstrating superior performance in identifying AA, AD, and AUD compared to existing methods.

Knowledge of vacant taxis’ passenger-searching behavior is of great social and economic interest to multiple applications, particularly in planning and operating on-demand mobility services. The inherent stochasticity and dynamic nature in both passenger demand and drivers’ decision-making pose challenges in capturing vacant taxis movements. This study proposes a novel deep clustering framework to comprehensively uncover passenger-searching strategies from extensive trajectory data. Specifically, multiple features from vacant searching trips are extracted and analogously defined as multi-channel images, where each channel corresponds to a specific feature. A novel deep image clustering approach is then proposed, integrating a feature representation module utilizing convolutional neural networks (ConvNets), a self-expression module for affinity learning, a spectral clustering module, and a classification module for self-supervision. An effective training procedure is also presented for the proposed deep clustering framework. Experiments demonstrate the effectiveness of the proposed approach against benchmark methods. Based on the clustering results, common and specific passenger searching strategies are further revealed. Specifically, our findings highlight the importance of individual’s contextual experience in explaining searching behavior and operational efficiency. Moreover, drivers cruising without clear searching strategy often exhibit lower performance, and some drivers may gamble with peers to increase their chances of picking up passengers. These results deliver important justifications for future studies and provide managerial implications to improve on-demand mobility.

Recommender systems have become a fundamental service in most E-Commerce platforms, in which the matching stage aims to retrieve potentially relevant candidate items to users for further ranking. Recently, some efforts on extracting multi-interests from user's historical behaviors have demonstrated superior performance. However, the historical behaviors are not noise-free due to the possible misclicks or disturbances. Existing works mainly overlook the fact that the interests of a user are not only reflected by the historical behaviors, but also inherently regulated by the profile information. Hence, we are interested in exploiting the benefit of user profile in multi-interest learning to enhance candidate matching performance. To this end, a user-aware multi-interest learning framework (named UMI) is proposed in this paper to exploit both user profile and behavior information for candidate matching. Specifically, UMI consists of two main components: dual-attention routing and interest refinement. In the dual-attention routing, we firstly introduce a user-guided attention network to identify the important historical items with respect to the user profile. Then, the resultant importance weights are leveraged via the dual-attentive capsule network to extract the user's multi-interests. Afterwards, the extracted interests are utilized to highlight the corresponding user profile features for interest refinement, such that different user profiles can be incorporated into interest learning for diverse user preference understanding. Besides, to improve the model's discriminative capacity, we further devise a harder-negatives strategy to support model optimization. Extensive experiments show that UMI significantly outperforms state-of-the-art multi-interest modeling alternatives. Currently, UMI has been successfully deployed at Taobao App in Alibaba, serving hundreds of millions of users.

Recommender system usually faces popularity bias. From the popularity distribution shift perspective, the normal paradigm trained on exposed items (most are hot items) identifies that recommending popular items more frequently can achieve lower loss, thus injecting popularity information into item property embedding, e.g., id embedding. From the long-tail distribution shift perspective, the sparse interactions of long-tail items lead to insufficient learning of them. The resultant distribution discrepancy between hot and long-tail items would not only inherit the bias, but also amplify the bias. Existing work addresses this issue with inverse propensity scoring (IPS) or causal embeddings. However, we argue that not all popularity biases mean bad effects, i.e., some items show higher popularity due to better quality or conform to current trends, which deserve more recommendations. Blindly seeking unbiased learning may inhibit high-quality or fashionable items. To make better use of the popularity bias, we propose a co-training disentangled domain adaptation network (CD$^2$AN), which can co-train both biased and unbiased models. Specifically, for popularity distribution shift, CD$^2$AN disentangles item property representation and popularity representation from item property embedding. For long-tail distribution shift, we introduce additional unexposed items (most are long-tail items) to align the distribution of hot and long-tail item property representations. Further, from the instances perspective, we carefully design the item similarity regularization to learn comprehensive item representation, which encourages item pairs with more effective co-occurrences patterns to have more similar item property representations. Based on offline evaluations and online A/B tests, we show that CD$^2$AN outperforms the existing debiased solutions. Currently, CD$^2$AN has been successfully deployed at Mobile Taobao App and handling major online traffic.

This study presents a robust deep reinforcement learning (RL) approach for real-time, network-wide holding control with transfer synchronization, considering stochastic passenger demand and vehicle running time during daily operations. The problem is formulated within a multi-agent RL framework where each active trip is considered as an agent, which not only interacts with the environment but also with other agents in the considered transit network. A specific learning procedure is developed to learn robust policies by introducing$maxmin$optimization into the learning objective. The agents are trained via deep deterministic policy gradient algorithm (DDPG) using an extended actor-critic framework with a joint action approximator. The effectiveness of the proposed approach is evaluated in a simulator, which is calibrated using data collected from a transit network in Twin Cities Minnesota, USA. The learned policy is compared with no control, rule-based control and the rolling horizon optimization control (RHOC). Computational results suggest that RL approach can significantly reduce the online computation time by about 50% compared with RHOC. In terms of policy performance, under deterministic scenario, the average waiting time of RL approach is 1.3% higher than the theoretical lower bound of average waiting time; under stochastic scenarios, RL approach could reduce as much as 18% average waiting time than RHOC, and the performance relative to RHOC improves when the level of system uncertainty increases. Evaluation under disrupted environment also suggests that the proposed RL method is more robust against short term uncertainties. The promising results in terms of both online computational efficiency and solution effectiveness suggest that the proposed RL method is a valid candidate for real-time transit control when the dynamics cannot be modeled perfectly with system uncertainties, as is the case for the network-wide transfer synchronization problem.