Hyperparameters optimization of convolutional neural network based on local autonomous competition harmony search algorithm

Liu Dongmei,Ouyang Haibin,Li Steven,장춘량,Zhan Zhi-Hui 한국CDE학회 2023 Journal of computational design and engineering Vol.10 No.4

Because of the good performance of convolutional neural network (CNN), it has been extensively used in many fields, such as image, speech, text, etc. However, it is easily affected by hyperparameters. How to effectively configure hyperparameters at a reasonable time to improve the performance of CNNs has always been a complex problem. To solve this problem, this paper proposes a method to automatically optimize CNN hyperparameters based on the local autonomous competitive harmony search (LACHS) algorithm. To avoid the influence of complicated parameter adjustment of LACHS algorithm on its performance, a parameter dynamic adjustment strategy is adopted, which makes the pitch adjustment probability PAR and step factor BW dynamically adjust according to the actual situation. To strengthen the fine search of neighborhood space and reduce the possibility of falling into local optima for a long time, an autonomous decision-making search strategy based on the optimal state is designed. To help the algorithm jump out of the local fitting situation, this paper proposes a local competition mechanism to make the new sound competes with the worst harmonic progression of local selection. In addition, an evaluation function is proposed, which integrates the training times and recognition accuracy. To achieve the purpose of saving the calculation cost without affecting the search result, it makes the training time for each model depending on the learning rate and batch size. In order to prove the feasibility of LACHS algorithm in configuring CNN superparameters, the classification of the Fashion-MNIST dataset and CIFAR10 dataset is tested. The comparison is made between CNN based on empirical configuration and CNN based on classical algorithms to optimize hyperparameters automatically. The results show that the performance of CNN based on the LACHS algorithm has been improved effectively, so this algorithm has certain advantages in hyperparametric optimization. In addition, this paper applies the LACHS algorithm to expression recognition. Experiments show that the performance of CNN optimized based on the LACHS algorithm is better than that of the same type of artificially designed CNN. Therefore, the method proposed in this paper is feasible in practical application.

Technical evaluation and optimization of biological aerated filter(BAF) treating micro-polluted source water at low temperature

Dongmei Liu,Xin Wang,Xiaobo Li,Aiwen Wang 대한환경공학회 2022 Environmental Engineering Research Vol.27 No.4

In this study, polyethylene and zeolite, as media in two-stage biological aerated filters (BAF), were evaluated for simultaneous removal of ammonia nitrogen and organics in micro-polluted water under low temperature conditions. For specific source water, the contaminants removal characteristics are evaluated by single-factor experiment to determine the suitable pH, air-water ratio (AWR) and hydraulic load ratio (HLR). Ammonia nitrogen and permanganate index in BAFs were promoted effectively with pH = 7~8, AWR = 1.0:1.0~2.0:1.0, HLR = 1~4 m³/(m²·h), low pollution loads and a temperature of 10°C. Based on the response surface analysis, the optimum parameters were as follows: pH of 7.94, AWR of 1.41:1.0 and HLR of 1.39 m³/(m²·h), for simultaneous removal of ammonia nitrogen and permanganate index, their removal rates were respectively 59.76% and 18.16%. BAF has a great advantage in emergency treatment of low temperature micro-polluted water, and response surface analysis can be used to optimize operating parameters.

Kinetic Resolution of Allylic Alcohol with Chiral BINOL-Based Alkoxides: A Combination of Experimental and Theoretical Studies

Liu, Yidong,Liu, Song,Li, Dongmei,Zhang, Nan,Peng, Lei,Ao, Jun,Song, Choong Eui,Lan, Yu,Yan, Hailong American Chemical Society 2019 JOURNAL OF THE AMERICAN CHEMICAL SOCIETY - Vol.141 No.2

<P>The development and characterization of enantioselective catalytic kinetic resolution of allylic alcohols through asymmetric isomerization with chiral BINOL derivatives-based alkoxides as bifunctional Brønsted base catalysts were described in the study. A number of chiral BINOL derivatives-based alkoxides were synthesized, and their structure-enantioselectivity correlation study in asymmetric isomerization identified a promising chiral Brønsted base catalyst, which afforded various chiral secondary allylic alcohols (ee up to 99%, <I>S</I> factor up to >200). In the mechanistic study, alkoxide species were identified as active species and the phenol group of BINOL largely affected the high reactivity and enantioselectivity via hydrogen bonding between the chiral Brønsted base catalyst and substrates. The strategy is the first successful synthesis strategy of various chiral secondary allylic alcohols through enantioselective transition-metal-free base-catalyzed isomerization. The applicability of the strategy had been demonstrated by the synthesis of the bioactive natural product (+)-veraguensin.</P> [FIG OMISSION]</BR>

Negative working capital management

Dongmei Wang,Na Liu 인하대학교 정석물류통상연구원 2009 인하대학교 정석물류통상연구원 학술대회 Vol.2009 No.10

Working capital is the most liquid and dynamic component of total assets. Working capital management, as the core content of short-term corporate finance, is vital to enterprises" survival and development and plays an important roel in maximizing shareholders" wealth. In this paper, we test the differences in profitability and risk levels from 2002 to 2006 between Chinese listed companies which pursue positive and negative working capital policies with the method of One-Way ANOVA. Furthermore, the first sample is eventually subdivided according to net commercial credits into two sub-ones, namely positive and negative commercial credits. Same analyses are done on above two sub-samples again. To find out how and what factors influence enterprises" profitability through different working capital polices, we do linear regression analyses. The results show: the proportion of Chinese listed companies which adopt the policy of negative working capital is steadily rising year by year; they have gained greater improvement in performance; however, relatively higher risk can not be ignored; companies can improve their performance by increasing the ratio of current assets to total assets (CAAR), the ratio of net cash inflows from operating activities to current liabilities (COACL), growth rate of total assets (GRA). In this paper, net commercial credits are firstly introduced into working capital research and as much attention is paid to risk indicators as to profitability ones. Such a practice is absolutely a meaningful try in this type of study. The conclusion of this paper not only supports the concept of "negative working capital" and its technical application, butl also gives useful suggestions to listed companies both in China and other countries.

Single nucleotide polymorphism-based analysis of the genetic structure of Liangshan pig population

Liu, Bin,Shen, Linyuan,Guo, Zhixian,Gan, Mailing,Chen, Ying,Yang, Runling,Niu, Lili,Jiang, Dongmei,Zhong, Zhijun,Li, Xuewei,Zhang, Shunhua,Zhu, Li Asian Australasian Association of Animal Productio 2021 Animal Bioscience Vol.34 No.7

Objective: To conserve and utilize the genetic resources of a traditional Chinese indigenous pig breed, Liangshan pig, we assessed the genetic diversity, genetic structure, and genetic distance in this study. Methods: We used 50K single nucleotide polymorphism (SNP) chip for SNP detection of 139 individuals in the Liangshan Pig Conservation Farm. Results: The genetically closed conserved population consisted of five overlapping generations, and the total effective content of the population (Ne) was 15. The whole population was divided into five boar families and one non-boar family. Among them, the effective size of each generation subpopulation continuously decreased. However, the proportion of polymorphic markers (P_N) first decreased and then increased. The average genetic distance of these 139 Liangshan pigs was 0.2823±0.0259, and the average genetic distance of the 14 boars was 0.2723±0.0384. Thus, it can be deduced that the genetic distance changed from generation to generation. In the conserved population, 983 runs of homozygosity (ROH) were detected, and the majority of ROH (80%) were within 100 Mb. The inbreeding coefficient calculated based on ROH showed an average value of 0.026 for the whole population. In addition, the inbreeding coefficient of each generation subpopulation initially increased and then decreased. In the pedigree of the whole conserved population, the error rate of paternal information was more than 11.35% while the maternal information was more than 2.13%. Conclusion: This molecular study of the population genetic structure of Liangshan pig showed loss of genetic diversity during the closed cross-generation reproduction process. It is necessary to improve the mating plan or introduce new outside blood to ensure long-term preservation of Liangshan pig.

NMAAP1 Expressed in BCG-Activated Macrophage Promotes M1 Macrophage Polarization

Liu, Qihui,Tian, Yuan,Zhao, Xiangfeng,Jing, Haifeng,Xie, Qi,Li, Peng,Li, Dong,Yan, Dongmei,Zhu, Xun Korean Society for Molecular and Cellular Biology 2015 Molecules and cells Vol.38 No.10

Macrophages are divided into two subpopulations: classically activated macrophages (M1) and alternatively activated macrophages (M2). BCG (Bacilli Calmette-$Gu{\acute{e}}rin$) activates disabled $na{\ddot{i}}ve$ macrophages to M1 macrophages, which act as inflammatory, microbicidal and tumoricidal cells through cell-cell contact and/or the release of soluble factors. Various transcription factors and signaling pathways are involved in the regulation of macrophage activation and polarization. We discovered that BCG-activated macrophages (BAM) expressed a new molecule, and we named it Novel Macrophage Activated Associated Protein 1 (NMAAP1). 1 The current study found that the overexpression of NMAAP1 in macrophages results in M1 polarization with increased expression levels of M1 genes, such as inducible nitric oxide synthase (iNOS), tumor necrosis factor alpha (TNF-${\alpha}$), Interleukin 6 (IL-6), Interleukin 12 (IL-12), Monocyte chemoattractant protein-1 (MCP-1) and Interleukin-1 beta (IL-$1{\beta}$), and decreased expression of some M2 genes, such as Kruppel-like factor 4 (KLF4) and suppressor of cytokine signaling 1 (SOCS1), but not other M2 genes, including arginase-1 (Arg-1), Interleukin (IL-10), transforming growth factor beta (TGF-${\beta}$) and found in inflammatory zone 1 (Fizz1). Moreover, NMAAP1 overexpression in the RAW264.7 cell line increased cytotoxicity against MCA207 tumor cells, which depends on increased inflammatory cytokines rather than cell-cell contact. NMAAP1 also substantially enhanced the phagocytic ability of macrophages, which implies that NMAAP1 promoted macrophage adhesive and clearance activities. Our results indicate that NMAAP1 is an essential molecule that modulates macrophages phenotype and plays an important role in macrophage tumoricidal functions.

Genetic homogeneity between populations of cotton bollworm from Xinjiang, China

Dongmei Wang,Xianming Yang,Haiqiang Li,Akedan Wuwaishi,Ruifeng Ding,Haobin Li,Hongsheng Pan,Jian Liu,Yao Xu,Yan-Hui Lu 한국응용곤충학회 2018 Journal of Asia-Pacific Entomology Vol.21 No.1

We studied the population structure of cotton bollworm (CBW), Helicoverpa armigera (Hübner) (Lepidoptera:Noctuidae), in Xinjiang, the largest cotton-growing region in China, using a fragment of cytochrome c oxidasesubunit I (COI) gene. Alignments of all 192 COI sequences revealed 28 haplotypes including 23 in southernXinjiang, 5 in eastern Xinjiang and 13 in northern Xinjiang. Negative and significant values of neutrality tests forthe Tajima's D and Fu's FS parameters, combined with the high values of haplotype diversity (Hd), low values ofnucleotide diversity (π) and a high number of low frequency haplotypes indicated a recent demographic expansionof Xinjiang CBW populations. Analysis of molecular variation (AMOVA) indicated low and non-significantgenetic structure, regardless of geographical scale or crop, with most of genetic variation occurringwithin local CBW populations. Pairwise FST analyses also indicated low genetic differentiation. This demographicevent and high gene flow could be responsible for the low genetic structure currently found. CBW populations inXinjiang need to be considered as one panmictic unit in its management, especially for the design of refuges todelay the development of resistance by this migratory pest to transgenic Bt (Bacillus thuringiensis) cotton.

Short Text Similarity Measure Based on Double Vector Space Model

Ying Liu,Dongmei Li,Cong Dai 보안공학연구지원센터 2016 International Journal of Database Theory and Appli Vol.9 No.10

Short text similarity measure is the basis of classification and duplicate checking of the short texts. Allowing for the insufficient consideration of the sentence semantic and structure information in similarity calculation between two short texts, we propose a novel method of short text similarity calculation based on double vector space model on the basis of traditional vector space model. Creatively transforming traditional vector space model into double vector space model. We utilize the numeral data link relations of Wikipedia to calculate semantic similarity between words, and calculate text structure similarity by dependency trees. Finally, we get the synthetic similarity by combining the semantic similar vector and structure similar vector. Our experiment results demonstrate that the proposed method has higher accuracy than other methods.

Gut Microbiota Dysbiosis Correlates With Long COVID-19 at One-Year After Discharge

Zhang Dongmei,Zhou Yaya,Ma Yanling,Chen Ping,Tang Jian,Yang Bohan,Li Hui,Liang Mengyuan,Xue YuE,Liu Yao,Zhang Jianchu,Wang Xiaorong 대한의학회 2023 Journal of Korean medical science Vol.38 No.15

Background: Long coronavirus disease 2019 (COVID-19) in recovered patients (RPs) is gradually recognized by more people. However, how long it will last and the underlining mechanism remains unclear. Methods: We conducted a prospective follow-up study to evaluate the long-term symptoms and clinical indices of RPs at one-year after discharge from Union Hospital, Wuhan, China between December 2020 to May 2021. We also performed the 16S rRNA sequencing of stool samples from RPs and healthy controls (HCs) and analyzed the correlation between the gut microbiota and long COVID-19. Results: In total, 187 RPs were enrolled, among them, 84 (44.9%) RPs reported long COVID-19 symptoms at one-year after discharge. The most common long-term symptoms were cardiopulmonary symptoms, including chest tightness after activity (39/187, 20.9%), palpitations on exercise (27/187, 14.4%), sputum (21/187, 11.2%), cough (15/187, 8.0%) and chest pain (13/187, 7.0%), followed by systemic symptoms including fatigue (34/187, 18.2%) and myalgia (20/187, 10.7%), and digestive symptoms including constipation (14/187, 7.5%), anorexia (13/187, 7.0%), and diarrhea (8/187, 4.3%). Sixty-six (35.9%) RPs presented either anxiety or depression (42/187 [22.8%] and 53/187 [28.8%] respectively), and the proportion of anxiety or depression in the long symptomatic group was significantly higher than that in the asymptomatic group (41/187 [50.6%] vs. 25/187 [24.3%]). Compared with the asymptomatic group, scores of all nine 36-Item Short Form General Health Survey domains were lower in the symptomatic group (all P < 0.05). One hundred thirty RPs and 32 HCs (non-severe acute respiratory syndrome coronavirus 2 infected subjects) performed fecal sample sequencing. Compared with HCs, symptomatic RPs had obvious gut microbiota dysbiosis including significantly reduced bacterial diversities and lower relative abundance of short-chain fatty acids (SCFAs)-producing salutary symbionts such as Eubacterium_hallii_group, Subdoligranulum, Ruminococcus, Dorea, Coprococcus, and Eubacterium_ventriosum_group. Meanwhile, the relative abundance of Eubacterium_hallii_group, Subdoligranulum, and Ruminococcus showed decreasing tendencies between HCs, the asymptomatic group, and the symptomatic group. Conclusion: This study demonstrated the presence of long COVID-19 which correlates with gut microbiota dysbiosis in RPs at one-year after discharge, indicating gut microbiota may play an important role in long COVID-19.

A New Multi-objective Evolutionary Algorithm for Inter-Cloud Service Composition

( Li Liu ),( Shuxian Gu ),( Dongmei Fu ),( Miao Zhang ),( Rajkumar Buyya ) 한국인터넷정보학회 2018 KSII Transactions on Internet and Information Syst Vol.12 No.1

Service composition in the Inter-Cloud raises new challenges that are caused by the different Quality of Service (QoS) requirements of the users, which are served by different geo-distributed Cloud providers. This paper aims to explore how to select and compose such services while considering how to reach high efficiency on cost and response time, low network latency, and high reliability across multiple Cloud providers. A new hybrid multi-objective evolutionary algorithm to perform the above task called LS-NSGA-II-DE is proposed, in which the differential evolution (DE) algorithm uses the adaptive mutation operator and crossover operator to replace the those of the Non-dominated Sorting Genetic Algorithm-II (NSGA-II) to get the better convergence and diversity. At the same time, a Local Search (LS) method is performed for the Non-dominated solution set F{1} in each generation to improve the distribution of the F{1}. The simulation results show that our proposed algorithm performs well in terms of the solution distribution and convergence, and in addition, the optimality ability and scalability are better compared with those of the other algorithms.