To address the challenges of catastrophic forgetting and limited generalization in Continual Few-shot Event Detection （CFED）， a new CFED model based on hierarchical adaptive fusion mechanism and category boundary distillation was proposed. Firstly， feature reconstruction was introduced by combining global average pooling with a learnable mapping to enhance the structural modeling of text representations and optimize feature distribution. Secondly， a hierarchical adaptive fusion mechanism was designed to dynamically integrate shallow， intermediate， and deep features from the pretrained model. Gaussian perturbation was introduced to improve feature robustness， and a self-attention mechanism was employed to achieve adaptive cross-layer feature weighted fusion. Finally， a category-boundary distillation strategy was proposed， which aligned the class distributions of old and new tasks using KL （Kullback-Leibler） divergence and refined the decision boundary features via cosine similarity， effectively mitigating knowledge forgetting. Experimental comparisons with 9 baseline models and the large language model GPT-3.5-Turbo were conducted on the MAVEN and ACE2005 datasets. On MAVEN， the proposed model achieved average F1 value improvements of 2.92 and 1.80 percentage points over the suboptimal model HANet （Hierarchical Augmentation Networks） across 5 subtasks under the 4-way 5-shot and 4-way 10-shot settings， respectively； on ACE2005， it outperformed the suboptimal models HANet and Combined Retrain by 1.83 and 2.00 percentage points across 5 subtasks under the 2-way 5-shot and 2-way 10-shot settings， respectively. Compared to GPT-3.5-Turbo， the proposed model achieved average F1 score improvements of 3.47 and 8.77 percentage points on MAVEN， and 4.47 and 2.39 percentage points on ACE2005 under 2-way 1-shot and 2-way 2-shot settings， respectively. The results demonstrate the superior performance of the proposed model.

Clustering is to divide a dataset into multiple clusters based on the similarity between samples. Most existing clustering methods face two challenges. On the one hand， when defining the similarity between samples， the spatial distribution structure of the samples is often not considered， making it difficult to construct a stable similarity matrix. On the other hand， the sample graph structure constructed by graph clustering is too complex and has high computational costs. To solve these two problems， a Multi-order Nearest Neighbor Graph Clustering algorithm by fusing transition probability matrix （MNNGC） was proposed. Firstly， the nearest neighbor relationship and spatial distribution structure of samples were comprehensively considered， the similarity defined by shared nearest neighbor was weighted for densification， and the densification affinity matrix between nodes was obtained. Secondly， by utilizing multi-order probability transition between nodes， the correlation degrees of non-adjacent nodes were predicted， and a stable inter-node affinity matrix was obtained by fusing the multi-order transition probability matrix. Then， to further enhance the local structure of the graph， the multi-order nearest neighbor graph of nodes was reconstructed， and hierarchically clustered. Finally， the edge node allocation strategy was optimized. Positioning experimental results show that MNNGC achieves the highest Accuracy （Acc） among comparison clustering algorithms on all the synthetic datasets and 8 UCI datasets. The Acc， Adjusted Mutual Information （AMI）， Adjusted Rand Index （ARI） and Fowlkes and Mallows Index （FMI） of MNNGC algorithm are improved by 38.6， 27.2， 45.4 and 35.1 percentage points， respectively， compared with Local Density Peaks-based Spectral Clustering （LDP-SC） algorithm.

In terms of the issue of reversible data hiding algorithm in videos, a novel algorithm based on histogram-pair method was proposed, which embedded data by selecting reasonable fluctuation value, freguercy range and area in I frames Discrete Cosine Transform (DCT) field, achieved high quality embedded MPEG-4 video. By embedding data in the macroblocks (8×8) in the optimum area, optimum frequency range in the macroblocks, and optimum DCT fluctuation, optimum reversible data hiding was completed. Higher Peak Signal-to-Noise Ratio (PSNR) was achieved in the experiments of 6 sequences videos. For akiyo, the PSNR of embedded I frame reached 45.33dB （1000b/frame），43.58dB （2000b/frame）和40.28dB (4000b/frame)。In the cases of high capacity, the increase of bit rate is relatively low, approximately 6% on average. The proposed method embedded data in DCT coefficient, achieves higher PSNR than the method based on DCT quantization table. The method embedded information in I frame beter than in B frame, which has formed relatively completed reversible data hiding method in video.

Software security measurement is critical to the development of software and improvement of software security. Based on the entropy and attack surface proposed by Manadhata et al. (MANADHATA P K, TAN K M C, MAXION R A, et al. An approach to measuring a system's attack surface, CMU-CS-07-146. Pittsburgh: Carnegie Mellon University, 2007; MANADHATA P K, WING J M. An attack surface metric. IEEE Transactions on Software Engineering, 2011, 37(3): 371-386), a method of software security measurement was used to assess the threat of the software's resources and provide the threat weight of these resources. Based on the threat weight, the attack surface metric was calculated for determining whether a software product is secure in design, or in what aspect the software product can be improved. The method is demonstrated in a case to show that, when using the method, the probable security threats can be found as early as possible to prevent from producing the software products that may have vulnerabilities, and the directions for the improvement of software security are pointed out clearly.

Most of the existing image data hiding methods have little payload and the visual effect is usually not good enough. The paper proposed a reversible hiding technique based on predictionerror and histogram pair. The gray value of a pixel was predicted using eight pixels around it in a multigray image. Then the predictionerror could be calculated. The hiding data were embedded into the predictionerror with the technique of histogram pair. An embedded threshold and an undulating threshold were selected during embedding,and adjusting the pair of thresholds achieved the best performance. The experimental results have demonstrated the proposed method obtains good performance both in embedding payload and visual quality.

The paper briefly introduced the concept of privacy preserving data mining technology and studied the application of decision tree classifier in this particular field. A decision tree classifier was applied and a scalar product protocol was added, so that the need of privacy preserving is satisfied as well as the advantage of decision tree is retained.