Retinal degenerative diseases such as age-related macular degeneration and retinitis pigmentosa cause severe vision impairment, while current electrical stimulation therapies are limited by poor spatial targeting precision. As a promising non-invasive alternative, the efficacy of temporal interference stimulation (TIS) for retinal targeting depends on optimized multi-electrode parameters. This study reconstructed a whole-head finite element model with detailed ocular structures and applied reinforcement learning (RL)-based multi-channel electrode parameter optimization to retinal stimulation. Systematic evaluation demonstrated that the focal precision of TIS improves with increasing channel numbers (consistent across all subject head models), with RL significantly outperforming conventional genetic algorithms (GA) and unsupervised neural networks (USNN) in focusing capability. Furthermore, by implementing the computationally intensive envelope calculation using the JAX framework, we achieved a nearly order-of-magnitude reduction in optimization time (to approx. 2 minutes per run on an RTX 4090D), significantly enhancing the practical feasibility of the proposed RL framework. This work provides a novel and computationally efficient methodology for precise non-invasive neuromodulation parameter optimization, applicable not only to retinal diseases but potentially to broader neurological conditions.

Existing literature has widely studied the impact of individual differences on ability-based social comparison in the workplace while the role of social media as a technical factor has not been fully understood. Drawing on social comparison theory, this study explored how social media affordances influence ability-based social comparison and the downstream impact on employees’ emotions and behaviors. We adopted the method of questionnaire survey and analyzed data from 239 participants. The results indicated that social media affordances have a positive impact on ability-based social comparison and the relationship is stronger for employees with low self-esteem. Ability-based social comparison facilitates envy and schadenfreude. Employees’ envy further exerts a negative impact by reducing altruistic behavior. This study contributes to the research on human-computer interaction and social comparison in the workplace, and provides guidance on how to maximize the value of social media in organizations.

This study investigates how herd behavior affects consumers’ consumption intention and continuous watching intention in live streaming e-commerce. We also consider how consumer–anchor interaction and consumer–consumer interaction moderate the effects of herd behavior on consumption intention and continuous watching intention. We specifically targeted users who watched Taobao live streaming to participate in our questionnaire, resulting in 296 valid responses. Our findings show that herd behavior positively affects consumption intention and continuous watching intention. Consumer–anchor interaction negatively moderates the link between herd behavior and both consumption intention and continuous watching intention. Consumer–consumer interaction positively moderates the relationship between herd behavior and consumption intention, but not the relationship between herd behavior and continuous watching intention. Furthermore, our findings show that herd behavior has a more pronounced impact on consumption intention than on continuous watching intention. Our results provide pertinent and beneficial implications for anchors, online merchants and platforms.

The surge in popularity of e-commerce live streaming is evident among sellers and consumers. However, a holistic examination of the influencing factors of consumers’ purchase intention from the perspective of social and technical systems is still limited. Utilizing the socio-technical systems theory, we investigate the effects of social and technical system factors on consumers’ flow experience and how flow affects consumer purchase intention in the live streaming setting. Furthermore, we identify how optimal stimulation level moderates the relationship between socio-technical system factors and flow. Empirical results (N = 355) indicate that flow can be strengthened through social system factors (financial bonds, social bonds) and technical system factors (visibility, guidance shopping), consequently affecting consumer purchase intention. In terms of the moderating effects, the optimal stimulation level positively moderates the relationships between social bonds, visibility, guidance shopping, and flow.

This paper constructs a risk reduction model based on signalling theory to investigate how the platform- and user-oriented signals affect perceived risks, which in turn affect users’ actual engagement and purchase behaviours in social commerce. The moderating effects of perceived effectiveness of social commerce institutional mechanisms (PESIM) on the relationships between signals and perceived risks are also considered. Longitudinal data from 226 users of Xiaohongshu were collected to test the proposed hypotheses. We identify signals oriented from platforms and users significantly influence users’ perceptions of commerce and participation risks. We also demonstrate that commerce and participation risks influence users’ actual engagement and purchase behaviours, respectively. Furthermore, we also verify that PESIM negatively moderates the relationship between signals and perceived risk.

Outlier detection is an important task in data mining, and many technologies for it have been explored in various applications. However, owing to the default assumption that outliers are not concentrated, unsupervised outlier detection may not correctly identify group anomalies with higher levels of density. Although high detection rates and optimal parameters can usually be achieved by using supervised outlier detection, obtaining a sufficient number of correct labels is a time-consuming task. To solve these problems, we focus on semi-supervised outlier detection with few identified anomalies and a large amount of unlabeled data. The task of semi-supervised outlier detection is first decomposed into the detection of discrete anomalies and that of partially identified group anomalies, and a distribution construction sub-module and a data augmentation sub-module are then proposed to identify them, respectively. In this way, the dual multiple generative adversarial networks (Dual-MGAN) that combine the two sub-modules can identify discrete as well as partially identified group anomalies. In addition, in view of the difficulty of determining the stop node of training, two evaluation indicators are introduced to evaluate the training status of the sub-GANs. Extensive experiments on synthetic and real-world data show that the proposed Dual-MGAN can significantly improve the accuracy of outlier detection, and the proposed evaluation indicators can reflect the training status of the sub-GANs.

A smart community provides various local services (i.e. smart community services [SCS]) to community residents through smart community platforms to improve their living environment and quality of life. The adoption of SCS by residents is critical to smart community initiatives, which is the concern of practitioners and researchers. However, few studies have empirically investigated factors affecting residents’willingness to adopt SCS. The present study empirically analysed how technological belief factors (i.e. perceived usefulness and enjoyment) and social influence factor (i.e. affective community commitment) influence SCS adoption intention. A total of 191 community residents in China were surveyed to test the research model. Results show that perceived usefulness, perceived enjoyment and affective community commitment are significant drivers of SCS adoption intention. Moreover, affective community commitment attenuates the impact of perceived enjoyment but enhances the effect of perceived usefulness on SCS adoption intention. This study enriches the literature on IT acceptance and offers practical suggestions for practitioners.

In speech synthesis with sparse training data, phonetic decision trees are frequently used for balance between model complexity and avail-able data. The traditional training procedure is that decision trees are constructed after parameters for each phones optimized in the EM al-gorithm. This paper proposes an iterative re-optimization algorithm in which the decision tree is re-learned after every iteration of the EM algorithm. The performance of the new procedure is compared with the original procedure by training parameters for MFCC and F0 features using an EDHMM model with data from The Boston Uni-versity Radio Speech corpus. A convergence proof is presented, and experimental tests demonstrate that iterative re-optimization gener-ates statistically significant test corpus log-likelihood improvements. Index Terms — speech synthesis, speech clustering, EM algo-rithm, decision tree 1.