Search

Ensemble-based debiasing methods have been shown effective in mitigating the reliance of classifiers on specific dataset bias, by exploiting the output of a bias-only model to adjust the learning target. In this paper, we focus on the bias-only model in these ensemble-based methods, which plays an important role but has not gained much attention in the existing literature. Theoretically, we prove that the debiasing performance can be damaged by inaccurate uncertainty estimations of the bias-only model. Empirically, we show that existing bias-only models fall short in producing accurate uncertainty estimations. Motivated by these findings, we propose to conduct calibration on the bias-only model, thus achieving a three-stage ensemble-based debiasing framework, including bias modeling, model calibrating, and debiasing. Experimental results on NLI and fact verification tasks show that our proposed three-stage debiasing framework consistently outperforms the traditional two-stage one in out-of-distribution accuracy.

Research of modeling infrared radiation characteristic of ocean background plays an important role in areas like marine remote sensing, prevention and control of ocean pollution, meteorological observation, and so on. In this paper, we build a three-D ocean surface model based on P-M wave spectrum, then set up the camera projection model. We use LOWTRAN 7 to calculate solar irradiance and sky background radiance, and then use thermal radiation theory to calculate the thermal radiation of the ocean itself and use bidirectional reflectance distribution function to calculate the reflection of the sea to the solar radiation and sky background radiance. Finally, with all the above radiation components considered, we generate the ocean background infrared simulation image.

The advanced aero engines have various features such as light weight, high efficiency and high power, so its rotor system is comparably weak, while the rotor often has to endure heavier weight loads. With loads of assembling and operating, the relative position between jointing interfaces will variate due to the relative slide of the interfaces and the changes of contact status. Dislocation of jointing interfaces can induce a deviation of the center of mass (CM) of the rotor system from its rotating axis, which implies an additional unbalance and may cause an engine vibration problem. In this article, a mechanism model is established to study the random influence of joints over the unbalance of rotor system. Then researchers use a typical aero-engine rotor as an example to investigate this topic. This study indicates that the additional unbalance of rotor system can be ascribed to the relative dislocation between jointing interfaces. The additional unbalance also shows a strong uncertainty due to the randomness of interfaces’ dislocation. Methods to control the technological parameters of joints should be utilized during the designing span of a rotor system in order to significantly depress the unbalance of rotor system and prevent engine from vibration effectively.

Cable domes has been widely adopted around the world. A new tensile steel strip cable dome is proposed and a force-finding method of approaching initial shape is introduced. It is easy to be constructed by tow-lifting method without many scaffolds and connect rigid materials directly. The force-finding method of approaching initial shape is effective for the proposed structure.