Multi-Label Hashing for Dependency Relations among Multiple Objectives.

Learning hash functions have been widely applied for large-scale image retrieval. Existing methods usually use CNNs to process an entire image at once, which is efficient for single-label images but not for multi-label images. First, these methods cannot fully exploit independent features of different objects in one image, resulting in some small object features with important information being ignored. Second, the methods cannot capture different semantic information from dependency relations among objects. Third, the existing methods ignore the impacts of imbalance between hard and easy training pairs, resulting in suboptimal hash codes. To address these issues, we propose a novel deep hashing method, termed multi-label hashing for dependency relations among multiple objectives (DRMH). We first utilize an object detection network to extract object feature representations to avoid ignoring small object features and then fuse object visual features with position features and further capture dependency relations among objects using a self-attention mechanism. In addition, we design a weighted pairwise hash loss to solve the imbalance problem between hard and easy training pairs. Extensive experiments are conducted on multi-label datasets and zero-shot datasets, and the proposed DRMH outperforms many state-of-the-art hashing methods with respect to different evaluation metrics.

Modification of the Maxwell-Wagner Heterogeneous Dielectric Model for Heterogeneous Polymers and Emulsions.

In heterogeneous polymers and emulsions, the volume fraction of the discrete phase and the frequency of electromagnetic waves affect the accuracy of the dielectric model. The integral method was used to modify the Maxwell-Wagner (M-W) heterogeneous dielectric theory, and a new model for the complex dielectric constant of polymers and emulsions was established. The experimental data were compared with the results of the M-W heterogeneous dielectric integral modification model and other theoretical models for different frequencies and volume fractions of the discrete phase. We discovered that with a decreasing volume fraction of the discrete phase, the dominant frequency range of the integral modification model expanded. When the volume fraction of the discrete phase is 10%, the dominant frequency range reaches 3 GHz. When the volume fraction of the discrete phase is 1%, the dominant frequency range reaches 4 GHz. When the volume fraction of the discrete phase is 0.06%, the dominant frequency range of the real part reaches 9.6 GHz, and the dominant frequency range of the imaginary part reaches 7.2 GHz. These results verify the advantages of the M-W modification model, which provides a theoretical basis to study the dielectric properties of polymers and emulsions, as well as for microwave measurement.

Dielectric properties of wet steam based on a double relaxation time model.

The last stages of most steam turbines operate in wet steam, resulting in water erosion of the rotor blades and the reduction of turbine efficiency. Accurate measurement of steam wetness is the key to ensure an efficient and stable operation of steam turbines. The equivalent complex permittivity model of wet steam was established by Maxwell-Wagner non-homogeneous dielectric theory, and the complex permittivity distribution of frequency and temperature changes of saturated water, dry saturated steam, and wet steam was derived. The measurement experiments verified the above properties of dry saturated steam and wet steam. The complex permittivity of the wet steam is similar to that for the dry saturated steam. The real part increases with increasing frequency and temperature. When the frequency is large or the temperature is low, the real part approaches 1. The imaginary part increases first and then decreases with the increase of frequency. In addition, with the increase of temperature, the imaginary part becomes larger. When the temperature is low, the imaginary part is close to 0, which is independent of the frequency.