To efficiently solve Multi-dimensional Knapsack Problem （MKP） using Ring Theory-based Evolutionary Algorithm （RTEA）， after analyzing the inadequacies of existing repair operators： RO1 （based on the pseudo-utility ratio of items’ overall resource consumption） and RO3 （based on the value density across individual resource dimensions）， a new weighted repair optimization operator named RO4 was proposed by integrating complementary strategy. Additionally， an inheritance strategy was introduced to improve the global evolutionary operator of RTEA， and a self-adaptive reverse mutation operator suitable for MKP was proposed on the basis of Logistic model， along with a new algorithm IRTEA-RO4 for solving MKP. To validate its efficiency， IRTEA-RO4 was tested on 114 internationally recognized MKP benchmark instances and compared with six state-of-the-art algorithms for solving MKP. Experimental results demonstrate that for small-scale MKP instances， IRTEA-RO4 achieves the highest solution accuracy and fastest computation speed； for large-scale MKP instances， IRTEA-RO4 outperforms the best results of the six existing algorithms by 21% to 125% in solution quality， while also exhibiting superior average performance， enhanced stability， and faster computational speed.

Drug synthesis reactions， especially asymmetric reactions， are the key components of modern pharmaceutical chemistry. Chemists have invested a lot in manpower and resources to identify various chemical reaction patterns in order to achieve efficient synthesis and asymmetric catalysis. The latest researches of quantum mechanical computing and machine learning algorithms in this field have proved the great potential of accurate virtual screening and learning the existing drug synthesis reaction data by computers. However， the existing methods only use few single-modal data， and can only use the common machine learning methods due to the limitation of not enough data. This hinders their universal application in a wider range of scenarios. Therefore， two screening models of drug synthesis reaction integrating multimodal data were proposed for virtual screening of reaction yield and enantioselectivity. At the same time， a 3D conformation descriptor based on Boltzmann distribution was also proposed to combine the 3D spatial information of molecules with quantum mechanical properties. These two multimodal data fusion models were trained and verified in two representative organic synthesis reactions （C-N cross coupling reaction and N， S-acetal formation）. The R²（R-squared） of the former is increased by more than 1 percentage point compared with those of the baseline methods in most data splitting， and the MAE（Mean Absolute Error） of the latter is decreased by more than 0.5 percentage points compared with those of the baseline methods in most data splitting. It can be seen that the models based on multimodal data fusion will bring good performance in different tasks of organic reaction screening.

The convergence speed of Artificial Glowworm Swarm Optimization (AGSO) algorithm declines, even falls into local minimums, when some glowworms gather in non whole extreme points or some glowworms wander around aimlessly. Concerning this problem, an AGSO algorithm based on adaptive t distribution mixed mutation was proposed. Adaptive t distribution mutation and optimization adjustment mutation was introduced into the AGSO algorithm to improve the diversity of glowworm swarm, and prevent the AGSO algorithm from falling into local minimums. Mutation control factor was defined. Combining history status information, the description of adaptive t distribution mixed mutation was given. The mutation method could enhance ability of global exploration and local development. The emulation results of representative test functions and many application examples show that the proposed algorithm is reliable and efficient. Meanwhile, this algorithm is better than tradition algorithm in terms of speed and precision for seeking the optimum.

In order to effectively resolve the “semantic gap” exists in image retrieval, this paper studied a new method for semantic annotation. Based on image segmentation, the method constructed image dictionary during the training phase, through analysis and description of color, texture and wavelet contour, established the two-stage annotation model combining comparison of wavelet contour and probability, it adopted corresponding method for different images by phases. Experiment indicates the method can significantly improve recall ratio and precision ratio, the maximum of precision is 23.6%, results prove that the model can understand image better also has good annotation effect and retrieval performance.