一种新的决策树选择性集成学习方法

个体学习器的差异度是集成学习中的关键因素。流行的集成学习算法如Bagging通过重取样技术产生个体学习器的差异度。选择性集成从集成学习算法产生的个体学习器中选择一部分来集成,结果表明比原集成更好。但如何选择学习器是个难题。使用Q统计量度量两个学习器的差异度,提出一种新的决策树选择性集成学习方法。与C4.5,Bagging方法相比,表现出很好的效果。     

基于遗传规划和集成学习的恶意软件检测

近年来恶意软件不断地发展变化,导致单一检测模型的准确率较低,使用集成学习组合多种模型可以提高检测效果,但集成模型中基学习器的准确性和多样性难以平衡。为此,提出一种基于遗传规划的集成模型生成方法,遗传规划可以将特征处理和构建集成模型两个阶段集成到单个程序树中,解决了传统恶意软件集成检测模型难以平衡个体准确率和多样性的问题。该方法以集成模型的恶意软件检出率作为种群进化依据,保证了基学习器的准确性;在构建集成模型时自动选择特征处理方法、分类算法和优化基学习器的超参数,通过输入属性扰动和算法参数扰动增加基学习器的多样性,根据优胜劣汰的思想进化生成具有高准确性和多样性的最优集成模型。在EMBER数据集上的结果表明,最优集成模型的检测准确率达到了98.88%;进一步的分析表明,该方法生成的模型具有较高的多样性和可解释性。     

基于Bagging的选择性聚类集成

使用集成学习技术来提高聚类性能.由于聚类使用的训练样本缺乏期望输出,与监督学习下的集成相比,在对个体学习器进行结合时更加困难.通过对不同的聚类结果进行配准,并基于互信息权进行个体学习器的选择,提出了基于Bagging的选择性聚类集成算法.实验表明,该算法能够有效地改善聚类结果.     

基于子图策略的选择性分类器集成算法

为了去除集成学习中的冗余个体,提出了一种基于子图选择个体的分类器集成算法。训练出一批分类器,利用个体以及个体间的差异性构造出一个带权的完全无向图;利用子图方法选择部分差异性大的个体参与集成。通过使用支持向量机作为基学习器,在多个分类数据集上进行了实验研究,并且与常用的集成方法Bagging和Adaboost进行了比较,结果该方法获得了较好的集成效果。     

基于SVM的网络入侵检测集成学习算法

互联网络中,计算机和设备随时受到恶意入侵的威胁,严重影响了网络的安全性。入侵行为升级快、隐蔽性强、随机性高,传统方法难以有效防范。针对这一问题,提出一种基于SVM的网络入侵检测集成学习算法,该算法利用SVM建立入侵检测基学习器,采用AdaBoost集成学习方法对基学习器迭代训练,生成最终的入侵检测模型,仿真实验表明了该算法的有效性。     

基于成对差异性度量的选择性集成方法

有效地产生泛化能力强、差异大的个体学习器,是集成学习算法的关键。为了提高学习器的差异性和精度,文中提出一种基于成对差异性度量的选择性集成方法。同时研究一种改进方法,进一步提高方法的运算速度,且支持并行计算。最后通过使用BP神经网络作为基学习器,在UCI数据集上进行实验,并与Bagging、基于遗传算法的选择性集成(GASEN)算法进行比较。实验结果表明,该改进算法在性能上与GASEN算法相近的前提下,训练速度得到大幅提高。     

基于基学习器差异度的层次化Bagging 集成修剪

本文主要目的是寻找到Bagging 的一种快速修剪方法,以缩小算法占用的存储空间、提高运算速度和 实现提高分类精度的潜力;还提出一种直接计算基学习器差异度的新选择性集成思想．选择出基学习器集合中对提 升其余基学习器差异度能力最强者进行删除,通过层次修剪来加速这一算法．在不影响性能的基础上,新算法能够 大幅度缩小Bagging 的集成规模;新算法还支持并行计算,其进行选择性集成的速度明显优于GASEN．本文还给出 了集成学习分类任务的误差上界     

基于稀疏表示剪枝集成建模的烧结终点位置智能预测

烧结终点位置(BTP)是烧结过程至关重要的参数, 直接决定着最终烧结矿的质量. 由于BTP难以直接在线 检测, 因此, 通过智能学习建模来实现BTP的在线预测并在此基础上进行操作参数调节对提高烧结矿质量具有重要 意义. 针对这一实际工程问题, 首先提出一种基于遗传优化的Wrapper特征选择方法, 可选取使后续预测建模性能最 优的特征组合; 在此基础上, 为了解决单一学习器容易过拟合的问题, 提出了基于随机权神经网络(RVFLNs)的稀疏 表示剪枝(SRP)集成建模算法, 即SRP-ERVFLNs算法. 所提算法采用建模速度快、泛化性能好的RVFLNs 作为个体 基学习器, 采用对基学习器基函数与隐层节点数等参数进行扰动的方式来增加集成学习子模型间的差异性; 同时, 为了进一步提高集成模型的泛化性能与计算效率, 引入稀疏表示剪枝算法, 实现对集成模型的高效剪枝; 最后, 将所 提算法用于烧结过程BTP的预测建模. 工业数据实验表明, 所提方法相比于其他方法具有更好的预测精度、泛化性 能和计算效率.     

基于半监督回归的选择性集成算法

为了提高小数据量的有标记样本问题中学习器的性能,结合半监督学习和选择性集成学习,提出了基于半监督回归的选择性集成算法SSRES。算法基于半监督学习的基本思想,同时使用有标记样本和未标记样本训练学习器从而减少对有标记样本的需求,使用选择性集成算法GRES对不同学习器进行适当的选择,并将选择的结果结合提高学习器的泛化能力。实验结果表明,在小数据量的有标记样本问题中,该算法能够有效地提高学习器的性能。     

基于进化集成学习的用户购买意向预测

在电子商务时代背景下,精准预测用户的购买意向已经成为提高销售效率和优化客户体验的关键因素。针对传统集成策略在模型设计阶段往往受人为因素限制的问题,构建了一种自适应进化集成学习模型用于预测用户的购买意向。该模型能够自适应地选择最优基学习器和元学习器,并融合基学习器的预测信息和特征间的差异性扩展特征维度,从而提高预测的准确性。此外,为进一步优化模型的预测效果,设计了一种二元自适应差分进化算法进行特征选择,旨在筛选出对预测结果有显著影响的特征。研究结果表明,与传统优化算法相比,二元自适应差分进化算法在全局搜索和特征选择方面表现优异。相较于六种常见的集成模型和DeepForest模型,所构建的进化集成模型在AUC值上分别提高了2.76%和2.72%,并且能够缓解数据不平衡所带来的影响。     

协作学习中基于协同过滤的学习资源推荐研究

符合学习者特征的学习资源对于提高协作学习效率具有重要的影响。但是传统的学习资源推荐,没有充分考虑学习者、学习资源的特征和高效的推荐算法。针对上述问题,提出了基于协同过滤的学习资源推荐算法,根据学习者学习特征、学习资源特征和学习者对学习资源历史评价信息,采用协同过滤推荐算法,实现学习资源推荐。首先,通过学习者特征和学习资源的评分,寻找相似学习者并计算学习资源预测评分,然后根据该评分值和学习资源与学习者匹配度推荐学习资源,从而为学习者推荐符合自己兴趣爱好最合适的学习资源。实验结果表明该算法在个性化学习资源推荐的准确性上优于传统算法。     

Ensemble cluster pruning via convex-concave programming

Ensemble learning is the process of aggregating the decisions of different learners/models. Fundamentally, the performance of the ensemble relies on the degree of accuracy in individual learner predictions and the degree of diversity among the learners. The trade-off between accuracy and diversity within the ensemble needs to be optimized to provide the best grouping of learners as it relates to their performance. In this optimization theory article, we propose a novel ensemble selection algorithm which, focusing specifically on clustering problems, selects the optimal subset of the ensemble that has both accurate and diverse models. Those ensemble selection algorithms work for a given number of the best learners within the subset prior to their selection. The cardinality of a subset of the ensemble changes the prediction accuracy. The proposed algorithm in this study determines both the number of best learners and also the best ones. We compared our prediction results to recent ensemble clustering selection algorithms by the number of cardinalities and best predictions, finding better and approximated results to the optimum solutions.     

面向不平衡样本的物联网入侵检测方法

随着设备的迭代,网络流量呈现指数级别的增长,针对各种应用的攻击行为越来越多,从流量层面识别并对这些攻击流量进行分类具有重要意义。同时,随着物联网设备的激增,针对这些设备的攻击行为也逐渐增多,造成的危害也越来越大。物联网入侵检测方法可以从这些海量的流量中识别出攻击流量,从流量层面保护物联网设备,阻断攻击行为。针对现阶段各类攻击流量检测准确率低以及样本不平衡问题,提出了基于重采样随机森林（RF,random forest）的入侵检测模型——Resample-RF,共包含3种具体算法：最优样本选择算法、基于信息熵的特征归并算法、多分类贪心转化算法。在物联网环境中,针对不平衡样本问题,提出最优样本选择算法,增加小样本所占权重,从而提高模型准确率;针对随机森林特征分裂效率不高的问题,提出基于信息熵的特征归并算法,提高模型运行效率;针对随机森林多分类精度不高的问题,提出多分类贪心转化算法,进一步提高准确率。在两个公开数据集上进行模型的检验,在 IoT-23 数据集上 F1 达到0.99,在Kaggle数据集上F1达到1.0,均具有显著效果。从实验结果中可知,提出的模型具有非常好的效果,能从海量流量中有效识别出攻击流量,较好地防范黑客对应用的攻击,保护物联网设备,从而保护用户。     

Intrusion Detection System for Big Data Analytics in IoT Environment

In the digital area, Internet of Things (IoT) and connected objects generate a huge quantity of data traffic which feeds big data analytic models to discover hidden patterns and detect abnormal traffic. Though IoT networks are popular and widely employed in real world applications, security in IoT networks remains a challenging problem. Conventional intrusion detection systems (IDS) cannot be employed in IoT networks owing to the limitations in resources and complexity. Therefore, this paper concentrates on the design of intelligent metaheuristic optimization based feature selection with deep learning (IMFSDL) based classification model, called IMFSDL-IDS for IoT networks. The proposed IMFSDL-IDS model involves data collection as the primary process utilizing the IoT devices and is preprocessed in two stages: data transformation and data normalization. To manage big data, Hadoop ecosystem is employed. Besides, the IMFSDL-IDS model includes a hill climbing with moth flame optimization (HCMFO) for feature subset selection to reduce the complexity and increase the overall detection efficiency. Moreover, the beetle antenna search (BAS) with variational autoencoder (VAE), called BAS-VAE technique is applied for the detection of intrusions in the feature reduced data. The BAS algorithm is integrated into the VAE to properly tune the parameters involved in it and thereby raises the classification performance. To validate the intrusion detection performance of the IMFSDL-IDS system, a set of experimentations were carried out on the standard IDS dataset and the results are investigated under distinct aspects. The resultant experimental values pointed out the betterment of the IMFSDL-IDS model over the compared models with the maximum accuracy 95.25% and 97.39% on the applied NSL-KDD and UNSW-NB15 dataset correspondingly.     

Community-organizing agent: An artificial intelligent system for building learning communities among large numbers of learners

Web-based (or online) learning provides an unprecedented flexibility and convenience to both learners and instructors. However, large online classes relying on instructor-centered presentations could tend to isolate many learners. The size of these classes and the wide dispersion of the learners make it challenging for instructors to interact with individual learners or to facilitate learner collaborations. Since extensive literature has confirmed that the substantial impact of learner interaction on learning outcomes, it is pedagogically critical to help distributed learners engage in community-based collaborative learning and to help individual learners improve their self-regulation. The E-learning lab of Shanghai Jiaotong University created an artificial intelligence system to help guide learners with similar interests into reasonably sized learning communities. The system uses a multi-agent mechanism to organize and reorganize supportive communities based on learners’ learning interests, experiences, and behaviors. Through effective award and exchange algorithms, learners with similar interests and experiences will form a community to support each others’ learning. Simulated experimental results indicate that these algorithms can improve the speed and efficiency in identifying and grouping homogeneous learners. Here, we will describe this system in detail and present its mechanism for organizing learning communities. We will conduct human experimentations in the near future to further perfect the system.     

基于神经网络的学习状态检测

对在线学习者注意力状态检测的方法大多基于眼睛闭合频率、头部偏转等特征,此类方法能够应对大多数情况,但针对学习者正视屏幕且视线落点处于屏幕上时出现的发呆、分神状态无法作出检测。针对此问题,提出了一种基于RNN的眼动分析算法RNN-EMA(RNN-EyeMovementAnalysis),该算法通过对序列眼动向量分析,预测学生学习行为,完成当前学习状态检测。实验表明,RNN-EMA算法能够对学习状态作出有效检测,且对比同类方法效果有所提升。     

Automatic detection of learning styles for an e-learning system

A desirable characteristic for an e-learning system is to provide the learner the most appropriate information based on his requirements and preferences. This can be achieved by capturing and utilizing the learner model. Learner models can be extracted based on personality factors like learning styles, behavioral factors like user’s browsing history and knowledge factors like user’s prior knowledge. In this paper, we address the problem of extracting the learner model based on Felder–Silverman learning style model. The target learners in this problem are the ones studying basic science. Using NBTree classification algorithm in conjunction with Binary Relevance classifier, the learners are classified based on their interests. Then, learners’ learning styles are detected using these classification results. Experimental results are also conducted to evaluate the performance of the proposed automated learner modeling approach. The results show that the match ratio between the obtained learner’s learning style using the proposed learner model and those obtained by the questionnaires traditionally used for learning style assessment is consistent for most of the dimensions of Felder–Silverman learning style.     

Artificial Algae Optimization with Deep Belief Network Enabled Ransomware Detection in IoT Environment

The Internet of Things (IoT) has gained more popularity in research because of its large-scale challenges and implementation. But security was the main concern when witnessing the fast development in its applications and size. It was a dreary task to independently set security systems in every IoT gadget and upgrade them according to the newer threats. Additionally, machine learning (ML) techniques optimally use a colossal volume of data generated by IoT devices. Deep Learning (DL) related systems were modelled for attack detection in IoT. But the current security systems address restricted attacks and can be utilized outdated datasets for evaluations. This study develops an Artificial Algae Optimization Algorithm with Optimal Deep Belief Network (AAA-ODBN) Enabled Ransomware Detection in an IoT environment. The presented AAA-ODBN technique mainly intends to recognize and categorize ransomware in the IoT environment. The presented AAA-ODBN technique follows a three-stage process: feature selection, classification, and parameter tuning. In the first stage, the AAA-ODBN technique uses AAA based feature selection (AAA-FS) technique to elect feature subsets. Secondly, the AAA-ODBN technique employs the DBN model for ransomware detection. At last, the dragonfly algorithm (DFA) is utilized for the hyperparameter tuning of the DBN technique. A sequence of simulations is implemented to demonstrate the improved performance of the AAA-ODBN algorithm. The experimental values indicate the significant outcome of the AAA-ODBN model over other models.