LSTMAtU-Net: A Precipitation Nowcasting Model Based on ECSA Module.

Precipitation nowcasting refers to the use of specific meteorological elements to predict precipitation in the next 0-2 h. Existing methods use radar echo maps and the Z-R relationship to directly predict future rainfall rates through deep learning methods, which are not physically constrained, but suffer from severe loss of predicted image details. This paper proposes a new model framework to effectively solve this problem, namely LSTMAtU-Net. It is based on the U-Net architecture, equipped with a Convolutional LSTM (ConvLSTM) unit with the vertical flow direction and depthwise-separable convolution, and we propose a new component, the Efficient Channel and Space Attention (ECSA) module. The ConvLSTM unit with the vertical flow direction memorizes temporal changes by extracting features from different levels of the convolutional layers, while the ECSA module innovatively integrates different structural information of each layer of U-Net into the channelwise attention mechanism to learn channel and spatial information, thereby enhancing attention to the details of precipitation images. The experimental results showed that the performance of the model on the test dataset was better than other examined models and improved the accuracy of medium- and high-intensity precipitation nowcasting.

Cascading Alignment for Unsupervised Domain-Adaptive DETR with Improved DeNoising Anchor Boxes.

Transformer-based object detection has recently attracted increasing interest and shown promising results. As one of the DETR-like models, DETR with improved denoising anchor boxes (DINO) produced superior performance on COCO val2017 and achieved a new state of the art. However, it often encounters challenges when applied to new scenarios where no annotated data is available, and the imaging conditions differ significantly. To alleviate this problem of domain shift, in this paper, unsupervised domain adaptive DINO via cascading alignment (CA-DINO) was proposed, which consists of attention-enhanced double discriminators (AEDD) and weak-restraints on category-level token (WROT). Specifically, AEDD is used to aggregate and align the local-global context from the feature representations of both domains while reducing the domain discrepancy before entering the transformer encoder and decoder. WROT extends Deep CORAL loss to adapt class tokens after embedding, minimizing the difference in second-order statistics between the source and target domain. Our approach is trained end to end, and experiments on two challenging benchmarks demonstrate the effectiveness of our method, which yields 41% relative improvement compared to baseline on the benchmark dataset Foggy Cityscapes, in particular.

A Dual-Population-Based NSGA-III for Constrained Many-Objective Optimization.

The main challenge for constrained many-objective optimization problems (CMaOPs) is how to achieve a balance between feasible and infeasible solutions. Most of the existing constrained many-objective evolutionary algorithms (CMaOEAs) are feasibility-driven, neglecting the maintenance of population convergence and diversity when dealing with conflicting objectives and constraints. This might lead to the population being stuck at some locally optimal or locally feasible regions. To alleviate the above challenges, we proposed a dual-population-based NSGA-III, named DP-NSGA-III, where the two populations exchange information through the offspring. The main population based on the NSGA-III solves CMaOPs and the auxiliary populations with different environment selection ignore the constraints. In addition, we designed an ε-constraint handling method in combination with NSGA-III, aiming to exploit the excellent infeasible solutions in the main population. The proposed DP-NSGA-III is compared with four state-of-the-art CMaOEAs on a series of benchmark problems. The experimental results show that the proposed evolutionary algorithm is highly competitive in solving CMaOPs.

Structure optimization of the two-dimensional off-lattice hydrophobic-hydrophilic model.

A two-dimensional off-lattice protein model with two species of monomers, hydrophobic and hydrophilic, was studied. Low-energy configurations in the model were optimized using the improved energy landscape paving (ELP+) method. In ELP+, the energy landscape paving (ELP) was first applied to search for the low-energy states. After the ELP led to the basins of the local energy minima, the additional degree-of-freedom of bond length was introduced, and the gradient descent method was then used to search for lower energy states near the local minima. Numerical results show that the proposed methods are quite effective for finding the ground states of proteins. A comparison between ELP+ and other methods is made.