针对现有基于端到端方面的情感分析(E2E-ABSA)方法研究中没有充分利用文本信息的不足,提出了一种基于BERT与融合词性、句法信息(lexical and syntactic information,LSI)的模型LSI-BERT。使用BERT嵌入层和TFM特征提取器来提取语义信息,...针对现有基于端到端方面的情感分析(E2E-ABSA)方法研究中没有充分利用文本信息的不足,提出了一种基于BERT与融合词性、句法信息(lexical and syntactic information,LSI)的模型LSI-BERT。使用BERT嵌入层和TFM特征提取器来提取语义信息,并通过工业级自然语言处理工具SpaCy提取词性信息,引入两个权重因子α和β对语义与词性信息进行融合;采用图注意网络(graph attention networks,GAT)根据句法依存树生成的邻接矩阵进行句法依存信息的提取;利用双流注意力网络针对句法依存信息和融合了词性信息的文本信息进行融合,使这两种信息实现更好的交互。实验结果表明,模型在三个常用基准数据集上的性能优于当前代表模型。展开更多
目的整合生物信息学挖掘并分析与基底型乳腺癌(basal-like breast cancer,BLBC)的预后相关的核心基因。方法首先,从GEO数据库中遴选与乳腺癌分子分型相关的数据集,数据处理后利用WGCNA筛选与BLBC相关的模块。然后,借助蛋白-蛋白互作(pro...目的整合生物信息学挖掘并分析与基底型乳腺癌(basal-like breast cancer,BLBC)的预后相关的核心基因。方法首先,从GEO数据库中遴选与乳腺癌分子分型相关的数据集,数据处理后利用WGCNA筛选与BLBC相关的模块。然后,借助蛋白-蛋白互作(protein-protein interaction,PPI)网络和cytohubba筛选出模块中差异最大的前10%基因作为候选基因,对候选基因进行生存分析和表达分析得到核心基因。最后,利用TIMER、TISIDB等生信工具探索核心基因表达和肿瘤免疫浸润、趋化因子及免疫调节剂的相关性并构建核心基因转录调控网络。结果利用WGCNA筛选出与BLBC相关的黑色模块中共891个基因,从差异性最大的80个候选基因中分析获得ESPL1和CCNB2两个核心基因。结果显示,两个核心基因与BLBC免疫细胞浸润有关,主要包括Th2细胞、CD8+T细胞、内皮细胞和肿瘤相关成纤维细胞。而且,核心基因表达水平与趋化因子、免疫刺激因子、免疫抑制因子及MHC分子相关。核心基因上游转录调控网络表明22种转录因子同时调控两个核心基因。结论ESPL1和CCNB2是BLBC的预后标志物且与肿瘤免疫相关。展开更多
In order to overcome shortcomings of traditional BP neural network,such as low study efficiency, slow convergence speed,easily trapped into local optimal solution,we proposed an improved BP neural network model based ...In order to overcome shortcomings of traditional BP neural network,such as low study efficiency, slow convergence speed,easily trapped into local optimal solution,we proposed an improved BP neural network model based on adaptive particle swarm optimization( PSO) algorithm. This algorithm adjusted the inertia weight coefficients and learning factors adaptively and therefore could be used to optimize the weights in the BP network. After establishing the improved PSO-BP( IPSO-BP) model,it was applied to solve fault diagnosis of rolling bearing. Wavelet denoising was selected to reduce the noise of the original vibration signals,and based on these vibration signals a wide set of features were used as the inputs in the neural network models. We demonstrate the effectiveness of the proposed approach by comparing with the traditional BP,PSO-BP and linear PSO-BP( LPSO-BP) algorithms. The experimental results show that IPSO-BP network outperforms other algorithms with faster convergence speed,lower errors,higher diagnostic accuracy and learning ability.展开更多
文摘目的整合生物信息学挖掘并分析与基底型乳腺癌(basal-like breast cancer,BLBC)的预后相关的核心基因。方法首先,从GEO数据库中遴选与乳腺癌分子分型相关的数据集,数据处理后利用WGCNA筛选与BLBC相关的模块。然后,借助蛋白-蛋白互作(protein-protein interaction,PPI)网络和cytohubba筛选出模块中差异最大的前10%基因作为候选基因,对候选基因进行生存分析和表达分析得到核心基因。最后,利用TIMER、TISIDB等生信工具探索核心基因表达和肿瘤免疫浸润、趋化因子及免疫调节剂的相关性并构建核心基因转录调控网络。结果利用WGCNA筛选出与BLBC相关的黑色模块中共891个基因,从差异性最大的80个候选基因中分析获得ESPL1和CCNB2两个核心基因。结果显示,两个核心基因与BLBC免疫细胞浸润有关,主要包括Th2细胞、CD8+T细胞、内皮细胞和肿瘤相关成纤维细胞。而且,核心基因表达水平与趋化因子、免疫刺激因子、免疫抑制因子及MHC分子相关。核心基因上游转录调控网络表明22种转录因子同时调控两个核心基因。结论ESPL1和CCNB2是BLBC的预后标志物且与肿瘤免疫相关。
基金Sponsored by the National Natural Science Foundation of China(Grant No.61174115 and 51104044)
文摘In order to overcome shortcomings of traditional BP neural network,such as low study efficiency, slow convergence speed,easily trapped into local optimal solution,we proposed an improved BP neural network model based on adaptive particle swarm optimization( PSO) algorithm. This algorithm adjusted the inertia weight coefficients and learning factors adaptively and therefore could be used to optimize the weights in the BP network. After establishing the improved PSO-BP( IPSO-BP) model,it was applied to solve fault diagnosis of rolling bearing. Wavelet denoising was selected to reduce the noise of the original vibration signals,and based on these vibration signals a wide set of features were used as the inputs in the neural network models. We demonstrate the effectiveness of the proposed approach by comparing with the traditional BP,PSO-BP and linear PSO-BP( LPSO-BP) algorithms. The experimental results show that IPSO-BP network outperforms other algorithms with faster convergence speed,lower errors,higher diagnostic accuracy and learning ability.
