Search

In the field of game designing, artificial intelligence is used to generate responsive, adaptive, or intelligent behaviors primarily in Non-Player-Characters (NPCs). There is a large demand for controlling game AI since a variety of players expect to be provided NPC opponents with appropriate difficulties to improve their game experience. However, to the best of our knowledge, a few works are focusing on this problem. In this paper, we firstly present a Reinforced Evolutionary Algorithm based on the Difficulty-Difference objective (REA-DD) to the DLAI problem, which combines reinforcement learning and evolutionary algorithms. REA-DD is able to generate the desired difficulty level of game AI accurately. Nonetheless, REA can only obtain a kind of game AI in each run. To improve efficiency, another algorithm based on Multi-objective Optimization is proposed, regarded as RMOEA-DD, which obtains DLAI after one run. Experiments on the game Pong from ALE and apply on a commercial game named The Ghost Story to show that our algorithms provide valid methods to the DLAI problem both in the term of controlling accuracy and efficiency.

Although current quality indicators of colonoscopy recommend 6 minutes as the minimum standard for withdrawal time (WT), the impact of a WT longer than 6 minutes on neoplasia detection is unclear.A multicenter randomized controlled trial involving 1027 patients was conducted from January 2018 to July 2019. Participants were randomly divided into a 9-minute (n = 514) and 6-minute (n = 513) WT group, and a timer was used to adjust the withdrawal speed. The primary outcome was the adenoma detection rate (ADR).Intention-to-treat analysis showed a significantly higher ADR in the 9-minute versus 6-minute WT group (36.6% vs. 27.1%, P = .001). Prolonging WT from 6 to 9 minutes significantly increased ADR of the proximal colon (21.4% vs. 11.9%, P.001) as well as of the less experienced colonoscopists (36.8% vs. 23.5%, P = .001). Improvements were also observed in the polyp detection rate (58.0% vs. 47.8%, P.001), and mean number of polyps and adenomas detected per colonoscopy (1.1 vs. 0.9, P = .002; 0.5 vs. 0.4, P = .008, respectively). The higher ADRs in 9-minute WT were also confirmed by the per-protocol (PP) analysis and subgroup analyses, with an increased rate of sessile serrated lesion detection in the 9-minute WT by PP analysis (4.0% vs. 1.3%, P = .04). Multivariate logistic regression demonstrated that the 9-minute WT was independently associated with increased ADR (P = .005).Prolonging WT from 6 to 9 minutes significantly improved ADR, especially in the proximal colon and for less experienced colonoscopists. A 9-minute WT benchmark should be considered as one of the quality indicators of colonoscopy. ClinicalTrials.gov (identifier, NCT03399045).

The technology of Covert Channel is often used for communications between computers and the Internet with high sensitivity or security levels. Presently, some research were carried out on covert channel using computer screen light radiation, speakers, electromagnetic leakage, etc. In this paper, the technology of SoundHammer is studied. It is a technical bridgeware that use acoustic waves to transmit data from sound device into an air-gapped network. Firstly, an idea was proposed for data transmission by covert channel through acoustic waves. Then, this method was validated by experiment and the risks of the air-gapped network were confirmed. Finally, some countermeasures for detecting and eliminating such covert channels were listed.

Personality is an important psychological construct accounting for individual differences in people. To reliably, validly, and efficiently recognize an individual's personality is a worthwhile goal; however, the traditional ways of personality assessment through self-report inventories or interviews conducted by psychologists are costly and less practical in social media domains, since they need the subjects to take active actions to cooperate. This paper proposes a method of big five personality recognition (PR) from microblog in Chinese language environments with a new machine learning paradigm named label distribution learning (LDL), which has never been previously reported to be used in PR. One hundred and thirteen features are extracted from 994 active Sina Weibo users' profiles and micro-blogs. Eight LDL algorithms and nine non-trivial conventional machine learning algorithms are adopted to train the big five personality traits prediction models. Experimental results show that two of the proposed LDL approaches outperform the others in predictive ability, and the most predictive one also achieves relatively higher running efficiency among all the algorithms.

Depression has become a serious global disease which can harm to mental health. Diagnosis of depression is a complicated process, and needs a great deal of specialist knowledge. A person’s social network behavior can reflect his/her real psychological. The traditional process of diagnosing depression requires not only close coordination of depressed people but also too much professional knowledge. And depressed people maybe have fewer and fewer social activities. In this paper, machine learning is used to diagnosis whether a person has depression. We use a SVM algorithm to create the depression diagnosis model based on short text. And we validate this model using a data set of Sina Micro-blog users’ Micro-blog content and depression label data gotten by certificate of diagnosis. Our work makes three important contributions. Firstly, we show how to diagnosis depression based on short text published on the social network. Secondly, we build diagnosis model of depression based on SVM. And last and thirdly, we give the experimental results that validate our method, the accuracy of whether or not depression can reach 93.47%.

Personality prediction on social network has become a hot topic. At present, most studies are using single-task classification/regression machine learning. However, this method ignores the potential association between multiple tasks. Also an ideal prediction result is difficult to achieve based on the small scale training data, since it is not easy to get a lot of social network data with personality label samples. In this paper, a robust multi-task learning method (RMTL) is proposed to predict Big-Five personality on Micro-blog. We aim to learn five tasks simultaneously by extracting and utilizing appropriate shared information among multiple tasks as well as identifying irrelevant tasks. For a set of Sina Micro-blog users’ information and personality labeled data retrieved by questionnaire, we validate the RMTL method by comparing it with 4 single-task learning methods and the mere multi-task learning. Our experiment demonstrates that the proposed RMTL can improve the precision rate, recall rate of the prediction and F value.

Recent years, the significant advantage has been taken by ceramic industrial from the strong evolution in automation. All production phases have been addressed through various technical innovations, with the exception of the final stage of the manufacturing process. When applying computer vision on ceramic product quality inspection, the different kind of product of ceramic makes it difficult, especially on ceramic bowls, which have a curved surface. It is difficult to analyze rapidness and catch the production rate. This paper proposed an approach for automatic surface imperfection inspection of ceramic products. The bowl’s curved surface unfolding model was discussed derived from helicoid unfolding method. Experiment shows that this method can be nicety.

Setting up an electronic examination system over internet is an important component of distance education. That system can be used to automatically evaluate students' performance. Nevertheless, that system is accessible to anyone at any places over internet; Different students present their examination at different time. This paper discusses a novel method for creating such system which could give different students different examination papers. The method is defined by two phases, first it pick up question fairly from question database by knowledge points, second it decide whether the question should be selected or not. This paper focuses on how to pick up question from question database by lottery schedule algorithms. Some experiments have been done on lab to test this approach.

For evaluating the performance of students in electronic education, it is important to set up an electronic examination system over internet. Nevertheless, the system is accessible to anyone at any places over internet; Different students present their examination at different time. Ideally, an electronic examination system should give different student different examination paper. This paper proposed a novel approach that an electronic examination system used to generate examination papers, which based on some fuzzy logic. Fuzzy similarity is used to select question from database, interval-valued fuzzy set logic is used to decide whether a question should add to paper or not. Some experiments have been done on lab to test this approach.