The vulnerability of deep neural networks to adversarial attacks has raised significant concerns about the security and reliability of artificial intelligence systems. Adversarial training is an effective approach to enhance adversarial robustness. To address the issue that existing methods adopt fixed adversarial sample generation strategies but neglect the importance of the adversarial sample generation phase for adversarial training， an adversarial training method was proposed based on adaptive attack strength. Firstly， the clean sample and the adversarial sample were input into the model to obtain the output. Then， the difference between the model outputs of the clean sample and the adversarial sample was calculated. Finally， the change of the difference compared with the previous moment was measured to automatically adjust the strength of the adversarial sample. Comprehensive experimental results on three benchmark datasets demonstrate that compared with the baseline method Adversarial Training with Projected Gradient Descent （PGD-AT）， the proposed method improves the robust precision under AA （AutoAttack） attack by 1.92， 1.50 and 3.35 percentage points on three benchmark datasets， respectively， and the proposed method outperforms the state-of-the-art defense method Adversarial Training with Learnable Attack Strategy （LAS-AT） in terms of robustness and natural accuracy. Furthermore， from the perspective of data augmentation， the proposed method can effectively address the problem of diminishing augmentation effect during adversarial training.

Patent price evaluation is an important part of intellectual property right transactions. When evaluating patent prices, the impact of the market, law, and technical dimensions on patent prices was not considered effectively by the existing methods. And the market factor of patent plays an important role in the evaluation of patent prices. Aiming at the above problem, an automatic patent price evaluation method based on recurrent neural network was proposed. In this method, based on the market approach, various other factors were considered comprehensively, and the Gated Recurrent Unit (GRU) neural network method was used to realize the automatic evaluation of patent prices. Example tests show that, with the qualitative evaluation results of experts as the benchmark, the average relative accuracy of the proposed method is 0.85. And this average relative accuracy of the proposed method is increased by 3.66%, 4.94% and 2.41% of the average relative accuracies of Analytic Hierarchy Process (AHP), rough set theory method and Back Propagation (BP) neural network method respectively.

More and more studies begin to focus on emotional conversation generation. However, the existing studies tend to focus only on emotional factors and ignore the relevance and diversity of topics in dialogues, as well as the emotional tendency closely related to topics, which may lead to the quality decline of generated responses. Therefore, a topic-expanded emotional conversation generation model that integrated topic information and emotional factors was proposed. Firstly, the conversation context was globally-encoded, the topic model was introduced to obtain the global topic words, and the external affective dictionary was used to obtain the global affective words in this model. Secondly, the topic words were expanded by semantic similarity and the topic-related affective words were extracted by dependency syntax analysis in the fusion module. Finally, the context, topic words and affective words were input into a decoder based on the attention mechanism to prompt the decoder to generate topic-related emotional responses. Experimental results show that the model can generate rich and emotion-related responses. Compared with the model Topic-Enhanced Emotional Conversation Generation(TE-ECG), the proposed model has an average increase of 16.3% and 15.4% in unigram diversity(distinct-1) and bigram diversity(distinct-2); and compared with Seq2SeqA(Sequence to Sequence model with Attention), the proposed model has an average increase of 26.7% and 28.7% in unigram diversity(distinct-1) and bigram diversity(distinct-2).

Aiming at the problem of low multi-task offloading efficiency in the “cloud+edge” hybrid environment composed of “central cloud server and multiple edge servers”， a task offloading method based on probabilistic performance awareness and evolutionary game theory was proposed. Firstly， in a “cloud + edge” hybrid environment composed of “central cloud server and multiple edge servers”， assuming that all the edge servers distributed in it had time-varying volatility performance， the historical performance data of edge cloud servers was probabilistically analyzed by a task offloading method based on probabilistic performance awareness and evolutionary game theory for obtaining the evolutionary game model. Then， an Evolutionary Stability Strategy （ESS） of service offloading was generated to guarantee that each user could offload tasks on the premise of high satisfaction rate. Simulation experiments were carried out based on the cloud edge resource locations dataset and the cloud service performance test dataset， the test and comparison of different methods were carried out on 24 continuous time windows. Experimental results show that， the proposed method is better than traditional task offloading methods such as Greedy algorithm， Genetic Algorithm （GA）， and Nash-based Game algorithm in many performance indexes. Compared with the three comparison methods， the proposed method has the average user satisfaction rate higher by 13.7%， 117.0%， 13.8% respectively， the average offloading time lower by 6.5%， 24.9%， 8.3% respectively， and the average monetary cost lower by 67.9%， 88.7%， 18.0% respectively.

In practical application,the patent quality evaluation is usually adopted by experts scoring or the quality evaluation index designed by the experts,so that the evaluation results are subjective and cannot be agreed by the both sides of the evaluation at the same time. In order to solve these problems,a deep learning patent quality evaluation method based on paper similarity calculation was proposed. Firstly,the papers were selected as the objective evaluation data,and the papers were used to calculate the similarity with the patent for augmented data. Then,a deep neural network was introduced to train the quality evaluation model,which was able to realize the map between the similarity of the paper and the quality of the patent to be evaluated. Finally,the quality evaluation model was used to access the patent quality. With perfect score of 100,the simulation results show that in different fields,compared to the corresponding expert evaluation result,the deviation of patent quality evaluation scores obtained by the proposed method is lower than 4,indicating that the proposed method has an effective patent quality evaluation ability.

Chain code is one kind of coding technology, which can represent the line, curve and region boundary with small data storage. In order to improve the compression efficiency of chain code, a new compression vertex chain code named Improved Orthogonal 3-Direction Vertex Chain Code (IO3DVCC) was proposed. The statistical characteristic of the Vertex Chain Code (VCC) and the directional characteristic of the OrThogonal 3-direction chain code (3OT) were combined in the proposed new chain code, 5 code values were totally set. The combination of 1, 3 and the combination of 3, 1 in VCC were merged and expressed by code 1. The expression of the code 2 was the same with the corresponding code value of VCC. The expression of code 3 was the same as the code value 2 of 3OT. Code 4 and code 5 corresponded to the two continuous code value 1 of IO3DVCC and eight continuous code values 2 of VCC respectively. Based on Huffman coding, the new chain code was the indefinite length coding. The code value probability, average expression ability, average length and efficiency of IO3DVCC, Enhanced Relative 8-Direction Freeman Chain Code (ERD8FCC), Arithmetic encoding Variable-length Relative 4-direction Freeman chain code (AVRF4), Arithmetic coding applied to 3OT chain code (Arith_3OT), Compressed VCC (CVCC), and Improved CVCC (ICVCC) were calculated aiming at the contour boundary of 100 images. The experimental results show that the efficiency of I3ODVCC is the highest. The total code number, total binary bit number, and compression ratio relative to the 8-Direction Freeman Chain Code (8DFCC) of three kinds of chain codes including IO3DVCC, Arith_3OT, and ICVCC were calculated aiming at the contour boundary of 20 randomly selected images. The experimental results demonstrate that the compression effect of IO3DVCC is the best.

Available face detection and evaluation standards are usually only a qualitative detection of the face existing, and have no strict norms for the quantitative description of the face location in videos.In addition, some researches such as video face replacement have higher requirements for the continuity of the face position in the video sequences. To solve these two problems, compared with the previous face detection algorithms and the face tracking evaluation standards, a quantitative detection standard of the human face position in the video was proposed, and a modified method of video face position detection was put forward. The initial face location was firstly detected in the target area by the improved Haar-Like cascade classifier; then the pyramid optical flow method was used to predict the position of the face, at the same time the forward-backward error detection mechanism was introduced to the self-checking of results, and finally the location of human face was determined. The experimental results show that the detection standard can give the evaluation of the quantitative description of the detection algorithm in the video face detection, and the proposed detection algorithm has a great improvement in the time consistency of face position in the detection results.

Mobile social network data has complex network structure, mutual label influence between nodes, variety of information including interactive information, location information, and other complex information. As a result, it brings many challenges to identify the characteristics of the user. In response to these challenges, a real mobile network was studied, the differences between the tagged users with different characteristics were extracted using statistical analysis, then the user's features of age and sex were recognized using relational Markov network prediction model. Analysis shows that the user of different age and sex has significant difference in call probability at different times, call entropy, distribution and discreteness of location information, gather degree in social networks, as well as binary and ternary interaction frequency. With these features, an approach for inferring the user's age and gender was put forward, which used the binary and ternary interaction relation group template, combined with the user's own temporal and spatial characteristics, and calculated the total joint probability distribution by relational Markov network. The experimental results show that the prediction accuracy of the proposed recognition model is at least 8% higher compared to the traditional classification methods, such as C4.5 decision tree, random forest, Logistic regression and Naive Bayes.

In order to improve the parameter optimization results of cognitive wireless network, an immune optimization based parameter adjustment algorithm was proposed. Engine parameter adjustment of cognitive wireless network is a multi-objective optimization problem. Intelligent optimization method is suitable for solving it. Immune clonal optimization is an effective intelligent optimization algorithm. The mutation probability affects the searching capabilities in immune optimization. Cloud droplets have randomness and stable tendency in normal cloud model, so an adaptive mutation probability adjustment method based on cloud model was proposed, and it was used in parameter optimization of cognitive radio networks. The simulation experiments were done to test the algorithm under multi-carrier system. The results show that, compared with relative algorithms, the proposed algorithm has better convergence, and the parameter adjustment results are consistent with the preferences for the objectives function. It can get optimal parameter results of cognitive engine.

This paper introduced the research works on all kinds of chain code used in image processing and pattern recognition and a new chain code named Improved Compressed Vertex Chain Code (ICVCC) was proposed based on Compressed Vertex Chain Code (CVCC). ICVCC added one code value compared with CVCC and adopted Huffman coding to encode each code value to achieve a set of chain code with unequal length. The expression ability per code, average length and efficiency as well as compression ratio with respect to 8-Directions Freeman Chain Code (8DFCC) were calculated respectively through the statistis a large number of images. The experimental results show that the efficiency of ICVCC proposed this paper is the highest and compression ratio is ideal.

The sentiment analysis has aroused the interest of many researchers in recent years，since the subjective texts are useful for many applications. Sentiment analysis is to mine and analyze the subjective text, aiming to acquire valuable knowledge and information. This paper surveyed the status of the art of Chinese sentiment analysis. Firstly, the technique was introduced in detail, according to different granularity levels, namely word, sentence, and document; and the research of product review and news review were presented respectively. Then evaluation and corpus for Chinese text sentiment analysis were introduced. The difficulty and trend of Chinese text sentiment analysis were concluded finally. This paper focuses on the major methods and key technologies in this field, making detailed analysis and comparison.

A transmission rate aware candidate forwarder selection and ordering algorithm based on expected transmission delay of nodes was proposed. It first separated opportunistic route forwarding into two components: the anycast forwarding from source node to its candidate forwarders set, and the remaining forwarding from that candidate forwarders set to destination, and then the shortest expected transmission delay of opportunistic routing was computed iteratively. Finally, candidate forwarders were selected and ordered according to the shortest expected transmission delay of nodes. The simulation results indicate that the proposed algorithm can improve the performance of opportunistic routing obviously.

A method for realtime and realistic rendering virtual agricultural scenes that include group of solar greenhouse was proposed in this paper. This method used viewdependent continuous LOD models to reduce the amount of polygons needed to be rendered actually, and a collision detection algorithm was implemented based on bounding box. The rendering speed of the scene was improved by using visibility culling techniques. In the constructed virtual agricultural scenes, large-scale plants stands were created rapidly by using geometric transformation, while shadow volume algorithm was implemented to render shadow in the scene. These significantly improved the realism of virtual agricultural scene. The simulation results demonstrate the proposed method can reduce the number of rendered elements effectively, and improves the rendering speed with high realism. This can meet the needs for realtime and interactive walkthrough in large scale virtual scenes.

Combining proxy signature and convertible authenticated encryption scheme together,a proxy convertible authenticated encryption scheme and a(t,n) threshold were proposed,which could make a proxy signer/at least t of n proxy signers delegating an original signer authenticatedly encrypt a message to a designated receiver.

The events correlation techniques in security integration management systems were introduced. A normal architecture of the correlation engine was introduced, and some discussions on the critical technologies and the main achievements in the field were put forward. The directions of the technology development were analyzed and evaluated, such as pattern obtainment, engine distribution and performance promotion. At last, a solution based on hierarchical rules to correlate events was presented.