Segmentation of pneumonia lesions from CT scans of COVID-19 patients is important for accurate diagnosis and follow-up. Deep learning has a potential to automate this task but requires a large set of high-quality annotations that are difficult to collect. Learning from noisy training labels that are easier to obtain has a potential to alleviate this problem. To this end, we propose a novel noise-robust framework to learn from noisy labels for the segmentation task. We first introduce a noise-robust Dice loss that is a generalization of Dice loss for segmentation and Mean Absolute Error (MAE) loss for robustness against noise, then propose a novel COVID-19 Pneumonia Lesion segmentation network (COPLE-Net) to better deal with the lesions with various scales and appearances. The noiserobust Dice loss and COPLE-Net are combined with an adaptive self-ensembling framework for training, where an Exponential Moving Average (EMA) of a student model is used as a teacher model that is adaptively updated by suppressing the contribution of the student to EMA when the student has a large training loss. The student

To add greater theoretical precision to a fundamental construct in social exchange theory – namely, Gouldner's ‘norm of reciprocity’, this study developed a measure of Sahlins' generalized, balanced, and negative reciprocity types and validated its psychometric quality in China. For a comprehensive construct validation of the new scale, we carried out three studies. After generating a pool of items, we used a panel of experts to classify items according to conceptual definitions of the three reciprocity types. Using factor analysis, the first study revealed a factor structure consistent with Sahlins' reciprocity typology. In the second study, confirmatory factor analysis replicated this factor structure as well as demonstrated that the reciprocity factors are distinct from each other and other social‐exchange constructs. In line with extant theories, the third study corroborated a nomological network relating reciprocity types to external constructs. Given this broad array of evidence for its construct validity, future researchers can employ this validated scale to investigate various forms of social exchange in Chinese work settings.

Given the assumption that most employers would like to gain the trust of their employees, what would initiate this trust? This study explores the joint role of the employee‐organization relationship (EOR) and supervisory support in initiating trust among middle managers. The results from a study of 545 middle managers in China show that both EOR and supervisory support are important in creating trust, with supervisory support having a stronger influence than EOR. Further, supervisors play a synergistic role by accentuating the positive influence of the mutual investment EOR approach and attenuating the negative influence of the quasi‐spot contract EOR with the managers. Results reinforce the importance of both formal structure and social processes in cultivating employee trust in the organization. We discuss implications of these findings for future research and human resource management practices. © 2008 Wiley Periodicals, Inc.

Translated and revised from Xinli Xuebao 心理学报 (Acta Psychologica Sinica), 2005, 37(6): 803-811 LI Chaoping ( )Abstract Using an open-ended questionnaire, this research firstly collects data from 249 managers and employees in various companies to identify behaviors and characteristics of transformational leadership in China. A content analysis reveals that transformational leadership includes eight specific categories in China. Then, transformational leadership questionnaire (TLQ) is developed in the second study by means of expert discussion. Exploratory factor analysis (EFA) of data from a sample of 431 employees shows that transformational leadership is a four-dimension construct in China, including moral modeling, charisma, articulate vision and individualized consideration. The third study further confirms TLQ's construct validity through confirmatory factor analysis (CFA) of data from another sample of 440 managers and employees. Internal consistency analyses and hierarchical regression analyses indicate that TLQ has sound reliability and concurrent validity.

This study examines the influence of resilience and transformational leadership on work engagement, and it investigates the mediating effect of positive affect. A total of 422 employees at a large IT company participated the survey. Participants completed established measures of resilience, transformational leadership, positive affect, and work engagement. The results indicate that resilience and transformational leadership are positively related to work engagement. Structural equation modeling analysis shows that positive affect partially mediates the relationships between resilience, transformational leadership, and work engagement. Theoretical contributions, practical implications, and future research directions are discussed.