Deep learning models are known to often learn features that spuriously correlate with the class label during training but are irrelevant to the prediction task. Existing methods typically address this issue by annotating potential spurious attributes, or filtering spurious features based on some empirical assumptions (e.g., simplicity of bias). However, these methods may yield unsatisfactory performance due to the intricate and elusive nature of spurious correlations in real-world data. In this paper, we propose a data-oriented approach1to mitigate the spurious correlation in deep learning models. We observe that samples that are influenced by spurious features tend to exhibit a dispersed distribution in the learned feature space. This allows us to identify the presence of spurious features. Subsequently, we obtain a bias-invariant representation by neutralizing the spurious features based on a simple grouping strategy. Then, we learn a feature transformation to eliminate the spurious features by aligning with this bias-invariant representation. Finally, we update the classifier by incorporating the learned feature transformation and obtain an unbiased model. By integrating the aforementioned identifying, neutralizing, eliminating and updating procedures, we build an effective pipeline for mitigating spurious correlation. Experiments on image and NLP debiasing benchmarks show an improvement in worst group accuracy of more than 20% compared to standard empirical risk minimization (ERM).

Cross-domain adaptive segmentation is a practical solution for the scenario that lacks expensive annotations or is inaccessible to ground truth. Prior works have tried to improve cross-domain adaptive segmentation with domain alignment, but most of them ignore the problem of training target deviation of distance-regularizing based domain alignment method. To address this, we propose a novel domain alignment mechanism that unifies the two optimization objectives, domain alignment, and segmentation performance, into one. In addition, existing methods are hard to apply under the source-free setting. We introduce a novel domain adaptive segmentation framework suitable for vanilla Unsupervised Domain Adaptation (UDA) and source-free UDA settings. Experiments show the proposed method outperforms competitive works with much more complicated mechanisms and achieves the state-of-the-art performance on both GTA→Cityscapes and Synthia→Cityscapes benchmarks. Our work can be easily added to existing methods and boost their performance.

The motivation of the paper is to find out the influence of demographic characteristic factors on employees’ career promotion in China’s state-owned enterprises and study the relationship between promotion mechanism and enterprise benefits. More than 6,500 pieces of human resource data from China’s state-owned enterprises were studied. Logistic regression is adopted to analyze the correlation between demographic characteristics and promotion. Meanwhile, different data mining methods are used to summarize the characteristics of promotion. On this basis, this study uses the principal-agent model of game theory to analyze the profits of employees and enterprises under different promotion mechanisms and demonstrates the conditions for enterprises to obtain the maximum benefit from the employees’ promotion. The results provide a theoretical reference for the assessment indicator selection of enterprise promotion and help executives better understand the impact of promotion mechanism on enterprises and employees. The application of results can reduce the information asymmetry in promotion incentive, prevent the emergence of adverse selection, and achieve a win-win situation for enterprises and employees.

By leveraging online social networks as an underlying infrastructure, Social Participatory Network (SPN) has been becoming a new paradigm of participatory sensing systems. However, a significant barrier to the widespread use of SPN applications is their vulnerability to various forms of malicious attacks. Such threats inhibit human participation and thus the viability of SPN systems in everyday use. To solve this problem, this paper proposes a trust evaluation framework for participants to encourage wider human participation in SPN. The proposal is based on the Tianjin University's own existing SPN system, named CRCS (ClassRoom Cloud System), which enables participants to use the cloud resources for online lessons or library study. It derives the trust value of participants by using entropy-weight method and data mining algorithms to deal with the behaviors data of participants. Our proposed solution can detect malicious participants easily, and more importantly, it outperforms other work for its low cost and simple deployment. For now, though our solution is based on a specified SPN system, we are confident that this solution is highly applicable to most other SPN systems.