基于弱监督预训练深度模型的微博情感分析

微博情感分析对于商业事务和政治选举等应用非常重要。传统的做法主要基于浅层机器学习模型,对人工提取的特征有较大的依赖,而微博情感特征往往难以提取。深度学习可以自动学习层次化的特征,并被用于解决情感分析问题。随着新的深度学习技术的提出,人们发现只要提供足够多的监督数据,就能训练出好的深度模型。然而,在微博情感分析中,通常监督数据都非常少。微博中广泛存在着弱监督数据。该文提出基于弱监督数据的“预训练—微调整”训练框架(distant pretrain-finetune),使用弱监督数据对深度模型进行预训练,然后使用监督数据进行微调整。这种做法的好处是可以利用弱监督数据学习到一个初始的模型,然后利用监督数据来进一步改善模型并克服弱监督数据存在的一些问题。我们在新浪微博数据上进行的实验表明,这种做法可以在监督数据较少的情况下使用深度学习,并取得比浅层模型更好的效果。     

Aerial scene classification via an ensemble extreme learning machine classifier based on discriminative hybrid convolutional neural networks features

Identifying a discriminative feature can effectively improve the classification performance of aerial scene classification. Deep convolutional neural networks (DCNN) have been widely used in aerial scene classification for its learning discriminative feature ability. The DCNN feature can be more discriminative by optimizing the training loss function and using transfer learning methods. To enhance the discriminative power of a DCNN feature, the improved loss functions of pretraining models are combined with a softmax loss function and a centre loss function. To further improve performance, in this article, we propose hybrid DCNN features for aerial scene classification. First, we use DCNN models with joint loss functions and transfer learning from pretrained deep DCNN models. Second, the dense DCNN features are extracted, and the discriminative hybrid features are created using linear connection. Finally, an ensemble extreme learning machine (EELM) classifier is adopted for classification due to its general superiority and low computational cost. Experimental results based on the three public benchmark data sets demonstrate that the hybrid features obtained using the proposed approach and classified by the EELM classifier can result in remarkable performance.     

Two-level hierarchical feature learning for image classification

In some image classification tasks, similarities among different categories are different and the samples are usually misclassified as highly similar categories. To distinguish highly similar categories, more specific features are required so that the classifier can improve the classification performance. In this paper, we propose a novel two-level hierarchical feature learning framework based on the deep convolutional neural network (CNN), which is simple and effective. First, the deep feature extractors of different levels are trained using the transfer learning method that fine-tunes the pre-trained deep CNN model toward the new target dataset. Second, the general feature extracted from all the categories and the specific feature extracted from highly similar categories are fused into a feature vector. Then the final feature representation is fed into a linear classifier. Finally, experiments using the Caltech-256, Oxford Flower-102, and Tasmania Coral Point Count (CPC) datasets demonstrate that the expression ability of the deep features resulting from two-level hierarchical feature learning is powerful. Our proposed method effectively increases the classification accuracy in comparison with flat multiple classification methods.     

基于视角-规则的深度TSK模糊分类器及其在多元癫痫脑电信号识别中的应用

在癫痫脑电信号分类检测中,传统机器学习方法分类效果不理想,深度学习模型虽然具有较好的特征学习优势,但其“黑盒”学习方式不具备可解释性,不能很好地应用于临床辅助诊断;并且,现有的多视角深度TSK模糊系统难以有效表征各视角特征之间的相关性.针对以上问题,提出一种基于视角-规则的深度Takagi-SugenoKang (TSK)模糊分类器(view-to-rule Takagi-Sugeno-Kang fuzzy classifier, VR-TSK-FC),并将其应用于多元癫痫脑电信号检测中.该算法在原始数据上构建前件规则以保证模型的可解释性,利用一维卷积神经网络(1-dimensional convolutional neural network, 1D-CNN)从多角度抓取多元脑电信号深度特征.每个模糊规则的后件部分分别采用一个视角的脑电信号深度特征作为其后件变量,视角-规则的学习方式提高了VR-TSK-FC表征能力.在Bonn和CHB-MIT数据集上, VR-TSK-FC算法模糊逻辑推理过程保证可解释的基础上达到了较好分类效果.     

A deep learning approach to the classification of 3D CAD models简

Model classification is essential to the management and reuse of 3D CAD models. Manual model classification is laborious and error prone. At the same time, the automatic classification methods are scarce due to the intrinsic complexity of 3D CAD models. In this paper, we propose an automatic 3D CAD model classification approach based on deep neural networks. According to prior knowledge of the CAD domain, features are selected and extracted from 3D CAD models first, and then pre-processed as high dimensional input vectors for category recognition. By analogy with the thinking process of engineers, a deep neural network classifier for 3D CAD models is constructed with the aid of deep learning techniques. To obtain an optimal solution, multiple strategies are appropriately chosen and applied in the training phase, which makes our classifier achieve better per-formance. We demonstrate the efficiency and effectiveness of our approach through experiments on 3D CAD model datasets.     

A co-boost framework for learning object categories from Google Images with 1st and 2nd order features

Conventional object recognition techniques rely heavily on manually annotated image datasets to achieve good performances. However, collecting high quality datasets is really laborious. The image search engines such as Google Images seem to provide quantities of object images. Unfortunately, a large portion of the search images are irrelevant. In this paper, we propose a semi-supervised framework for learning visual categories from Google Images. We exploit a co-training algorithm, the CoBoost algorithm, and integrate it with two kinds of features, the 1st and 2nd order features, which define bag of words representation and spatial relationship between local features, respectively. We create two boosting classifiers based on the 1st and 2nd order features in the training, during which one classifier provides labels for the other. The 2nd order features are generated dynamically rather than extracted exhaustively to avoid high computation. An active learning technique is also introduced to further improve the performance. Experimental results show that the object models learned from Google Images by our method are competitive with the state-of-the-art unsupervised approaches and some supervised techniques on the standard benchmark datasets.     

基于深度卷积神经网络的脑电信号情感识别

为了点对点自动学习脑电信号（Electroencephalogram,EEG）空间与时间维度上的情感相关特征,提高脑电信号情感识别的准确率,基于DEAP数据集中EEG信号的时域、频域特征及其组合特征,提出一种基于卷积神经网络（Convolution Neural Network,CNN）模型的EEG情感特征学习与分类算法。采用包括集成决策树、支持向量机、线性判别分析和贝叶斯线性判别分析算法在内的浅层机器学习模型与CNN深度学习模型对DEAP数据集进行效价和唤醒度两个维度上的情感分类实验。实验结果表明,在效价和唤醒度两个维度上,深度CNN模型在时域和频域组合特征上均取得了目前最好的两类识别性能,在效价维度上比最佳的传统分类器集成决策树模型提高了3.58%,在唤醒度上比集成决策树模型的最好性能提高了3.29%。     

Multi-scale features for heartbeat classification using directed acyclic graph CNN

ABSTRACT

A new architecture of deep neural networks, directed acyclic graph convolutional neural networks (DAG-CNNs), is used to classify heartbeats from electrocardiogram (ECG) signals into different subject-based classes. DAG-CNNs not only fuse the feature extraction and classification stages of the ECG classification into a single automated learning procedure, but also utilized multi-scale features and perform score-level fusion of multiple classifiers automatically. Therefore, DAG-CNN negates the necessity to extract hand-crafted features. In most of the current approaches, only the high level features which extracted by the last layer of CNN are used. Instead of performing feature level fusion manually and feeding the results into a classifier, the proposed multi-scale system can automatically learn different level of features, combine them and predict the output label. The results over the MIT-BIH arrhythmia benchmarks database demonstrate that the proposed system achieves a superior classification performance compared to most of the state-of-the-art methods.     

On Taxonomy and Evaluation of Feature Selection‐Based Learning Classifier System Ensemble Approaches for Data Mining Problems

Ensemble methods aim at combining multiple learning machines to improve the efficacy in a learning task in terms of prediction accuracy, scalability, and other measures. These methods have been applied to evolutionary machine learning techniques including learning classifier systems (LCSs). In this article, we first propose a conceptual framework that allows us to appropriately categorize ensemble‐based methods for fair comparison and highlights the gaps in the corresponding literature. The framework is generic and consists of three sequential stages: a pre‐gate stage concerned with data preparation; the member stage to account for the types of learning machines used to build the ensemble; and a post‐gate stage concerned with the methods to combine ensemble output. A taxonomy of LCSs‐based ensembles is then presented using this framework. The article then focuses on comparing LCS ensembles that use feature selection in the pre‐gate stage. An evaluation methodology is proposed to systematically analyze the performance of these methods. Specifically, random feature sampling and rough set feature selection‐based LCS ensemble methods are compared. Experimental results show that the rough set‐based approach performs significantly better than the random subspace method in terms of classification accuracy in problems with high numbers of irrelevant features. The performance of the two approaches are comparable in problems with high numbers of redundant features.     

Sentiment analysis of tweets using a unified convolutional neural network‐long short‐term memory network model

Sentiment analysis focuses on identifying and classifying the sentiments expressed in text messages and reviews. Social networks like Twitter, Facebook, and Instagram generate heaps of data filled with sentiments, and the analysis of such data is very fruitful when trying to improve the quality of both products and services alike. Classic machine learning techniques have a limited capability to efficiently analyze such large amounts of data and produce precise results; they are thus supported by deep learning models to achieve higher accuracy. This study proposes a combination of convolutional neural network and long short‐term memory (CNN‐LSTM) deep network for performing sentiment analysis on Twitter datasets. The performance of the proposed model is analyzed with machine learning classifiers, including the support vector classifier, random forest (RF), stochastic gradient descent (SGD), logistic regression, a voting classifier (VC) of RF and SGD, and state‐of‐the‐art classifier models. Furthermore, two feature extraction methods (term frequency‐inverse document frequency and word2vec) are also investigated to determine their impact on prediction accuracy. Three datasets (US airline sentiments, women's e‐commerce clothing reviews, and hate speech) are utilized to evaluate the performance of the proposed model. Experiment results demonstrate that the CNN‐LSTM achieves higher accuracy than those of other classifiers.     

Phonetic-enriched text representation for Chinese sentiment analysis with reinforcement learning

The Chinese pronunciation system offers two characteristics that distinguish it from other languages: deep phonemic orthography and intonation variations. In this paper, we hypothesize that these two important properties can play a major role in Chinese sentiment analysis. In particular, we propose two effective features to encode phonetic information and, hence, fuse it with textual information. With this hypothesis, we propose Disambiguate Intonation for Sentiment Analysis (DISA), a network that we develop based on the principles of reinforcement learning. DISA disambiguates intonations for each Chinese character (pinyin) and, hence, learns precise phonetic representations. We also fuse phonetic features with textual and visual features to further improve performance. Experimental results on five different Chinese sentiment analysis datasets show that the inclusion of phonetic features significantly and consistently improves the performance of textual and visual representations and surpasses the state-of-the-art Chinese character-level representations.     

An Ensemble Based Approach for Sentiment Classification in Asian Regional Language

In today’s digital world, millions of individuals are linked to one another via the Internet and social media. This opens up new avenues for information exchange with others. Sentiment analysis (SA) has gotten a lot of attention during the last decade. We analyse the challenges of Sentiment Analysis (SA) in one of the Asian regional languages known as Marathi in this study by providing a benchmark setup in which we first produced an annotated dataset composed of Marathi text acquired from microblogging websites such as Twitter. We also choose domain experts to manually annotate Marathi microblogging posts with positive, negative, and neutral polarity. In addition, to show the efficient use of the annotated dataset, an ensemble-based model for sentiment analysis was created. In contrast to others machine learning classifier, we achieved better performance in terms of accuracy for ensemble classifier with 10-fold cross-validation (cv), outcomes as 97.77%, f-score is 97.89%.     

DF-Net: Deep fusion network for multi-source vessel segmentation

Accurate retinal vessel segmentation is very challenging. Recently, the deep learning based method has greatly improved performance. However, the non-vascular structures usually harm the performance and some low contrast small vessels are hard to be detected after several down-sampling operations. To solve these problems, we design a deep fusion network (DF-Net) including multiscale fusion, feature fusion and classifier fusion for multi-source vessel image segmentation. The multiscale fusion module allows the network to detect blood vessels with different scales. The feature fusion module fuses deep features with vessel responses extracted from a Frangi filter to obtain a compact yet domain invariant feature representation. The classifier fusion module provides the network more supervision. DF-Net also predicts the parameter of the Frangi filter to avoid manually picking the best parameters. The learned Frangi filter enhances the feature map of the multiscale network and restores the edge information loss caused by down-sampling operations. The proposed end-to-end network is easy to train and the inference time for one image is 41ms on a GPU. The model outperforms state-of-the-art methods and achieves the accuracy of 96.14%, 97.04%, 98.02% from three publicly available fundus image datasets DRIVE, STARE, CHASEDB1, respectively. The code is available at https://github.com/y406539259/DF-Net.     

融合全局时序和局部空间特征的伪造人脸视频检测方法

近年来,深度学习在人工智能领域表现出优异的性能。基于深度学习的人脸生成和操纵技术已经能够合成逼真的伪造人脸视频,也被称作深度伪造,让人眼难辨真假。然而,这些伪造人脸视频可能会给社会带来巨大的潜在威胁,比如被用来制作政治虚假新闻,从而引发政治暴力或干扰正常选举等。因此,亟需研发对应的检测方法来主动发现伪造人脸视频。现有的方法在制作伪造人脸视频时,容易在空间上和时序上留下一些细微的伪造痕迹,比如纹理和颜色上的扭曲或脸部的闪烁等。主流的检测方法同样采用深度学习,可以被划分为两类,即基于视频帧的方法和基于视频片段的方法。前者采用卷积神经网络（Convolutional Neural Network,CNN）发现单个视频帧中的空间伪造痕迹,后者则结合循环神经网络（Recurrent Neural Network,RNN）捕捉视频帧之间的时序伪造痕迹。这些方法都是基于图像的全局信息进行决策,然而伪造痕迹一般存在于五官的局部区域。因而本文提出了一个统一的伪造人脸视频检测框架,利用全局时序特征和局部空间特征发现伪造人脸视频。该框架由图像特征提取模块、全局时序特征分类模块和局部空间特征分类模块组成。在FaceForensics++数据集上的实验结果表明,本文所提出的方法比之前的方法具有更好的检测效果。     

Bias and stability of single variable classifiers for feature ranking and selection

Feature rankings are often used for supervised dimension reduction especially when discriminating power of each feature is of interest, dimensionality of dataset is extremely high, or computational power is limited to perform more complicated methods. In practice, it is recommended to start dimension reduction via simple methods such as feature rankings before applying more complex approaches. Single variable classifier (SVC) ranking is a feature ranking based on the predictive performance of a classifier built using only a single feature. While benefiting from capabilities of classifiers, this ranking method is not as computationally intensive as wrappers. In this paper, we report the results of an extensive study on the bias and stability of such feature ranking method. We study whether the classifiers influence the SVC rankings or the discriminative power of features themselves has a dominant impact on the final rankings. We show the common intuition of using the same classifier for feature ranking and final classification does not always result in the best prediction performance. We then study if heterogeneous classifiers ensemble approaches provide more unbiased rankings and if they improve final classification performance. Furthermore, we calculate an empirical prediction performance loss for using the same classifier in SVC feature ranking and final classification from the optimal choices.     

Semisupervised Encrypted Traffic Identification Based on Auxiliary Classification Generative Adversarial Network

The rapidly increasing popularity of mobile devices has changed the methods with which people access various network services and increased network traffic markedly. Over the past few decades, network traffic identification has been a research hotspot in the field of network management and security monitoring. However, as more network services use encryption technology, network traffic identification faces many challenges. Although classic machine learning methods can solve many problems that cannot be solved by port- and payload-based methods, manually extract features that are frequently updated is time-consuming and labor-intensive. Deep learning has good automatic feature learning capabilities and is an ideal method for network traffic identification, particularly encrypted traffic identification; Existing recognition methods based on deep learning primarily use supervised learning methods and rely on many labeled samples. However, in real scenarios, labeled samples are often difficult to obtain. This paper adjusts the structure of the auxiliary classification generation adversarial network (ACGAN) so that it can use unlabeled samples for training, and use the wasserstein distance instead of the original cross entropy as the loss function to achieve semisupervised learning. Experimental results show that the identification accuracy of ISCX and USTC data sets using the proposed method yields markedly better performance when the number of labeled samples is small compared to that of convolutional neural network (CNN) based classifier.     

Sentiment analysis in textual,visual and multimodal inputs using recurrent neural networks

Social networking platforms have witnessed tremendous growth of textual, visual, audio, and mix-mode contents for expressing the views or opinions. Henceforth, Sentiment Analysis (SA) and Emotion Detection (ED) of various social networking posts, blogs, and conversation are very useful and informative for mining the right opinions on different issues, entities, or aspects. The various statistical and probabilistic models based on lexical and machine learning approaches have been employed for these tasks. The emphasis was given to the improvement in the contemporary tools, techniques, models, and approaches, are reflected in majority of the literature. With the recent developments in deep neural networks, various deep learning models are being heavily experimented for the accuracy enhancement in the aforementioned tasks. Recurrent Neural Network (RNN) and its architectural variants such as Long Short Term Memory (LSTM) and Gated Recurrent Unit (GRU) comprise an important category of deep neural networks, basically adapted for features extraction in the temporal and sequential inputs. Input to SA and related tasks may be visual, textual, audio, or any combination of these, consisting of an inherent sequentially, we critically investigate the role of sequential deep neural networks in sentiment analysis of multimodal data. Specifically, we present an extensive review over the applicability, challenges, issues, and approaches for textual, visual, and multimodal SA using RNN and its architectural variants.

     

Improved AdaBoost-based image retrieval with relevance feedback via paired feature learning

Boost learning algorithm, such as AdaBoost, has been widely used in a variety of applications in multimedia and computer vision. Relevance feedback-based image retrieval has been formulated as a classification problem with a small number of training samples. Several machine learning techniques have been applied to this problem recently. In this paper, we propose a novel paired feature AdaBoost learning system for relevance feedback-based image retrieval. To facilitate density estimation in our feature learning method, we propose an ID3-like balance tree quantization method to preserve most discriminative information. By using paired feature combination, we map all training samples obtained in the relevance feedback process onto paired feature spaces and employ the AdaBoost algorithm to select a few feature pairs with best discrimination capabilities in the corresponding paired feature spaces. In the AdaBoost algorithm, we employ Bayesian classification to replace the traditional binary weak classifiers to enhance their classification power, thus producing a stronger classifier. Experimental results on content-based image retrieval (CBIR) show superior performance of the proposed system compared to some previous methods.     

Ensemble sparse supervised model for bearing fault diagnosis in smart manufacturing

Machinery fault diagnosis is of great significance to improve the reliability of smart manufacturing. Deep learning based fault diagnosis methods have achieved great success. However, the features extracted by different models may vary resulting in ambiguous representation of the data, and even wasted time with manually selecting the optimal hyperparameters. To solve the problems, this paper proposes a new framework named Ensemble Sparse Supervised Model (ESSM), in which a typical deep learning model is treated as two phases of feature learning and model learning. In the feature learning phase, the original data is represented to be a feature matrix as non-redundant as possible by applying sparse filtering. Then, the feature matrix is fed into the model learning phase. Regularization, dropout and rectified linear unit (ReLU) are used in the model's neurons and layers to build a sparse deep neural network. Finally, the output of the sparse deep neural network provides feedback to the first phase to obtain better sparse features. In the proposed method, hyperparameters need to be pre-specified and a python library of talos is employed to finish the process automatically. The proposed method is verified using the bearing data provided by Case Western Reserve University. The result demonstrates that the proposed method can capture the effective pattern of data with the help of sparse constraints and simultaneously provide convenience for the operators with assuring performance.