Search

With the rapid development of the field of artificial intelligence, China is paying more and more attention to the cultivation of professional talents related to artificial intelligence. As an emerging cross-cutting discipline, our university has built a curriculum system for the major of intelligent science and technology, which is a collaborative platform of "industry, science, education and innovation" for the training of talents for new industries and new engineering disciplines. It also provides reference and inspiration for other majors in our university to implement the new engineering construction.

In recent years, with regard to the training of engineering talents, the phenomenon that theoretical knowledge and practical skills can not be well combined has become increasingly prominent, and it is urgent to reform and innovate the training mode and standards of engineering talents. At the same time, the reform of training mode should adhere to the development needs and problem-driven of national talents. This paper intends to study the training standards of new engineering talents, take coping with changes and shaping the future as the construction concept, take inheritance and innovation, cross-integration, coordination and sharing as the main ways, take training high-quality engineering and technical applied talents in the field of artificial intelligence as the goal, take the development needs of modern industry as the guidance, and combine with the discipline characteristics of the school. This paper studies the innovative construction mode of "Intelligent Science and Technology" specialty, and explores the establishment of "five integrations" training mechanism of enterprises and universities integration, science and education integration, universities integration, three-innovation integration and disciplines integration.

Mitochondrial thymidine kinase 2 (TK2) is an essential enzyme for mitochondrial dNTP synthesis in many tissues. Deficiency in TK2 activity causes devastating mitochondrial diseases. Here we investigated several residues involved in substrate binding and catalysis. We showed that mutations of Gln-110 and Glu-133 affected Mg2+ and ATP binding, and thus are crucial for TK2 function. Furthermore, mutations of Gln-110 and Tyr-141 altered the kinetic behavior, suggesting their involvement in substrate binding through conformational changes. Since the 3 D structure of TK2 is still unknown, and thus, the identification of key amino acids for TK2 function may help to explain how TK2 mutations cause mitochondrial diseases.

Clear-cell renal cell carcinoma is one of the most common tumors disagnosed, with nearly one third of patients diagnosed with metastatic ccRCC. Although an increasing number of studies has revealed that piwi-interacting RNAs are aberrantly expressed in diverse types of cancers, few of them explored the detailed molecular mechanism of piRNAs in carcinogenesis, particularly in ccRCC. In this study, differentially expressed piRNAs associated with ccRCC were selected by using piRNA-sequencing combined with TCGA data analysis, and piR-57125 was identified. PiR-57125 was found remarkably downregulated in ccRCC samples. Functionally, knockdown of piR-57125 promoted migration and invasion of ccRCC, while overexpression of piR-57125 suppressed ccRCC metastasis. In vivo lung metastasis model also confirmed the same results. CCL3 was identified as the direct target of piR-57125 which could potentially reverse the inhibition effect of piR-57125 in ccRCC metastasis. Further study revealed that piR-57125 modulated ccRCC metastasis through the AKT/ERK pathway. These data indicate that piR-57125 restrains ccRCC metastasis by directly targeting CCL3 and inhibiting the AKT/ERK pathway, and could be a potential therapeutic target for ccRCC.

Chronic sclerosing sialadenitis of the submandibular gland (also known as Küttner tumor) is characterized by concomitant swelling of the submandibular glands secondary to strong lymphocytic infiltration and fibrosis. The pathogenesis of this disease has been unclear, but it is associated with immune disorders. ADAMTS18 is a member of the ADAMTS superfamily of extracellular proteinases. In this study, we showed that Adamts18 is highly expressed in submandibular salivary gland (SMG) during embryonic development and decreases but is retained in adult SMG tissue in mice. Adamts18 deficiency led to reduced cleft formation and epithelial branching in embryonic SMG before embryonic day 15.5 in mice. No significant histologic changes in the later stages of branching or the morphology of SMG were detected in Adamts18−/− mice. However, Adamts18 deficiency causes spontaneous SMG fibrogenesis and fibrosis in adult mice. At 8 wk of age, Adamts18−/− mice began to manifest the first signs of pathologic changes of mild fibrosis and CD11b+ cell infiltration in SMG tissues. At ≥8 mo, all male and female Adamts18−/− mice developed unilateral or bilateral SMG scleroma that is similar to patients with chronic sclerosing sialadenitis of the submandibular gland. Adamts18−/− mice also showed secretory dysfunction and severe dental caries. Histologically, SMG scleroma is characterized by progressive periductal fibrosis, acinar atrophy, irregular duct ectasis, and dense infiltration of IgG-positive plasma cells. A significant infiltration of CD4+ T lymphocytes and CD11b+ monocytes and macrophages was also detected in the SMG scleroma of Adamts18−/− mice. The levels of TGF-β1, IL-6, and IL-33 were significantly increased in Adamts18−/− SMGs, which induces chronic inflammation and myofibroblast activation, ultimately leading to fibrosis. This study indicates that Adamts18 regulates the early branching morphogenesis of embryonic SMG and plays a role in protecting from spontaneous SMG fibrogenesis via modulating local inflammation, autoimmune reaction, and myofibroblast activation in adult mice.

Colletotrichum siamense, a member of Colletotrichum gloeosporioides complex species, is the primary pathogen causing rubber anthracnose, which leads to significant economic loss in natural rubber production. Velvet family proteins are fungal-specific proteins and play an essential role in regulating development and secondary metabolism. In this study, we characterized two velvet proteins CsVosA and CsVelB in C. siamense as the orthologs of VosA and VelB in Aspergillus nidulans. CsVosA is located in the nucleus, and CsVelB displays a localization in both the nucleus and the cytoplasm. Deleting CsvosA or CsvelB results in a slow growth rate, and the CsvelB-knockout mutants also exhibit low mycelial density. CsVosA and CsVelB are involved in regulating chitin metabolism and distribution, leading to the variation in the cell wall integrity of C. siamense. Furthermore, disruption of CsvosA or CsvelB can decrease conidial production and viability, and the ΔCsvosA and ΔCsvelB mutants also lose the ability to produce fruiting bodies. Pathogenicity assays show that deleting CsvosA or CsvelB can lower the virulence, and the two velvet genes are essential for the full virulence of C. siamense. Based on the results of the yeast two-hybrid analysis and bimolecular fluorescence complementation assays, CsVosA can interact with CsVelB and form the complex CsVosA-CsVelB in the conidia of C. siamense, which may play essential roles in maintaining the cell wall integrity and conidial viability. In addition, CsVelB is also involved in regulating melanin production of C. siamense. In conclusion, CsVosA and CsVelB regulate vegetative growth, cell wall integrity, asexual/sexual sporulation, conidial viability and virulence in C. siamense.

The heterogeneity of pore structure in coal reservoir is extremely complex. In this paper, mercury intrusion porosimetry (MIP) and liquid nitrogen adsorption (LNA) were used to describe pore characteristics of macro-coal components, and the fractal characteristics of pores and their relationship with adsorption and desorption were discussed. The findings revealed that there were obvious differences in pore characteristics of different macro-coal components at different pore sizes. The total pore volume of vitrain and durain was equivalent, and the total specific surface area was larger than durain, indicating that the micropores in vitrain were more developed, while the macropores in durain were more developed, indicating that the specific surface area was smaller. Fractal results indicated that the pore structure of coal was more complex with the increase of pore diameter. The Da1 and Da2 of vitrain and durain were affected by larger specific surface area and pore volume. The Ds of vitrain increased first and then decreased with the content of vitrinite, while that of durain was the opposite. The relationship between the adsorption capacity of vitrain and fractal dimension Da1 was binomial distribution, and it was positively correlated with Da2. The adsorption capacity of durain samples increased first and then decreased with Da1. With the increase of fractal dimension Ds, the theoretical desorption rate and recovery rate of durain had a downward trend, that is, the more complex the pore structure, the poor the desorption efficiency.

Nerves are important pathological elements of the microenvironment of tumors, including those in pancreatic, colon and rectal, prostate, head and neck, and breast cancers. Recent studies have associated perineural invasion with tumor progression and poor outcomes. In turn, tumors drive the reprogramming of neurons to recruit new nerve fibers. Therefore, the crosstalk between nerves and tumors is the hot topic and trend in current cancer investigations. Herein, we reviewed recent studies presenting direct supporting evidences for a better understanding of nerve–tumor interactions.

This study described electroencephalogram (EEG) parameters in children under general anesthesia, which could monitor patient-specific brain responses to anesthetics and assess the effects of anesthesia. The objective was to detect the patient state index (PSI) and associated factors. We analyzed EEG parameters in patients in the age range 1 to 36 months. Patients were stratified into 2 groups as those aged 1 to 12 months and 13 to 36 months. Sixty-two patients were involved. Spectral edge frequency (SEF), PSI, and blood pressure were lower, and burst suppression rate (BSR) and heart rate were higher in the 1 to 12 months group. The SEF was associated with PSI in both groups. Age and blood pressure were positively associated with PSI, and BSR was negatively related to PSI in children under 1 year of age. Blood pressure was not associated with PSI in the 13 to 36 months age group. We found that the PSI levels did not accurately assess the depth of anesthesia in children under 1 year of age.

With the application and development of Internet technology, network traffic is growing rapidly, and the situation of network security is becoming more and more serious. As an important way to protect network security, abnormal traffic detection has been paid more and more attention. In this paper, the uncertainty of the samples in the abnormal traffic detection dataset is studied. Combining the three-way decision idea with the random forest algorithm, a three-way selection random forest optimization model for abnormal traffic detection is proposed. Firstly, the three-way decision idea is integrated into the random selection process of feature attributes, and the attribute importance based on decision boundary entropy is calculated. The feature attributes are divided into the normal domain, abnormal domain, and uncertain domain, and the three-way attribute random selection rules are designed to randomly select the feature attributes that conform to the rules from different domains. Secondly, the classifier evaluation function is constructed by combining pure accuracy and diversity, and the anomaly traffic detection base classifier with a high evaluation value is selected for integration to eliminate the unstable factors caused by randomness in the process of base classifier generation. Thirdly, the optimal node weight combination of the base classifier is obtained by iterative calculation of the gray wolf optimization algorithm to further improve the prediction effect and robustness of the model. Finally, the model is applied to the abnormal traffic detection dataset. The experimental results show that the prediction accuracy of the three-way selection random forest optimization model on CIC-IDS2017, KDDCUP99, and NSLKDD datasets is 96.1%, 95.2%, and 95.3%, respectively, which has a better detection effect than other machine learning algorithms.

Aiming at the problems of the dynamic increase in data in real life and that the naive Bayes (NB) classifier only accepts or rejects the sample processing results, resulting in a high error rate when dealing with uncertain data, this paper combines three-way decision and incremental learning, and a new three-way incremental naive Bayes classifier (3WD-INB) is proposed. First, the NB classifier is established, and the distribution fitting is carried out according to the minimum residual sum of squares (RSS) for continuous data, so that 3WD-INB can process both discrete data and continuous data, then carry out an incremental learning operation, select the samples with higher data quality according to the confidence of the samples in the incremental training set for incremental learning, solve the problem of data dynamics and filter the poor samples. Then we construct the 3WD-INB classifier and determine the classification rules of the positive, negative and boundary domains of the 3WD-INB classifier, so that the three-way classification of samples can be realized and better decisions can be made when dealing with uncertain data. Finally, five discrete data and five continuous data are selected for comparative experimental analysis with traditional classification methods. The results show that 3WD-INB has high accuracy and recall rate on different types of datasets, and the classification performance is also relatively stable.

In recent years, deep learning models have been more and more widely used in various fields and have become a research hotspot for various tasks in artificial intelligence, but there are significant limitations in non-convex optimization problems. As a model training strategy for non-convex optimization, curriculum learning advocates that models learn in the order of easier to more difficult data, mimicking the basic idea of gradual human learning as they learn curriculum. This strategy has been widely used in the fields of computer vision, natural language processing, and reinforcement learning; it can effectively solve the non-convex optimization problem and improve the generalization ability and convergence speed of models. This paper first introduces the application of curriculum learning at three major levels: data, task, and model, and summarizes the evaluators designed using curriculum learning methods in various domains, including difficulty evaluators, training schedulers, and loss evaluators, which correspond to the three stages of difficulty evaluation, training schedule, and loss evaluation in the application of curriculum learning to model training. We also discuss how to choose an appropriate evaluation system and the differences between terms used in different types of research. Finally, we summarize five methods similar to curriculum learning in the field of machine learning and provide a summary and outlook of the curriculum learning evaluation system.

Metabolism reprograming is a hallmark of cancer and plays an important role in tumor progression. The aberrant metabolism in renal cell carcinoma (RCC) leads to accumulation of the oncometabolite l‐2‐hydroxyglurate (L‐2HG). L‐2HG has been reported to inhibit the activity of some α‐ketoglutarate‐dependent dioxygenases such as TET enzymes, which mediate epigenetic alteration, including DNA and histone demethylation. However, the detailed functions of L‐2HG in renal cell carcinoma have not been investigated thoroughly. In our study, we found that L‐2HG was significantly elevated in tumor tissues compared to adjacent tissues. Furthermore, we demonstrated that L‐2HG promoted vasculogenic mimicry (VM) in renal cancer cell lines through reducing the expression of PHLDB2. A mechanism study revealed that activation of the ERK1/2 pathway was involved in L‐2HG‐induced VM formation. In conclusion, these findings highlighted the pathogenic link between L‐2HG and VM and suggested a novel therapeutic target for RCC., What's new? Metabolic reprograming, a hallmark of cancer, influences tumor progression. In the case of renal cell carcinoma (RCC) specifically, progression appears to be facilitated by the oncometabolite L‐2‐hydroxyglurate (L‐2HG), though underlying mechanisms remain enigmatic. Here, the authors investigated the ability of L‐2HG in RCC to promote vasculogenic mimicry (VM), in which aggressive cancer cells form vessel‐like networks that support tumor growth. Analyses of RCC patient tissues revealed elevated L‐2HG levels, wherein tumor cells with greater L‐2HG levels exhibited more VM structures. TCGA data and high‐throughput sequencing analyses further show that L‐2HG contributes to VM formation via reduction of PHLDB2 levels.