卷积神经网络下的Twitter文本情感分析

随着社交网络的日益普及,基于Twitter文本的情感分析成为近年来的研究热点。Twitter文本中蕴含的情感倾向对于挖掘用户需求和对重大事件的预测具有重要意义。但由于Twitter文本短小和用户自身行为存在随意性等特点,再加之现有的情感分类方法大都基于手工制作的文本特征,难以挖掘文本中隐含的深层语义特征,因此难以提高情感分类性能。本文提出了一种基于卷积神经网络的Twitter文本情感分类模型。该模型利用word2vec方法初始化文本词向量,并采用CNN模型学习文本中的深层语义信息,从而挖掘Twitter文本的情感倾向。实验结果表明,采用该模型能够取得82.3%的召回率,比传统分类方法的分类性能有显著提高。     

用户-兴趣点耦合关系的兴趣点推荐方法

在基于位置的社交网络(LBSNs)中,如何利用用户和兴趣点的属性（或特征）之间的耦合关系,为用户做出准确的兴趣点推荐是当前的研究热点。现有的矩阵分解推荐方法利用用户对兴趣点的评分进行推荐,但评级矩阵通常非常稀疏,并且没有考虑用户和兴趣点在各自属性方面的耦合关系。本文提出了一种基于深度神经网络的兴趣点推荐框架,首先采用K-means算法对兴趣点按地理位置进行聚类,使位置相近的兴趣点聚为一类;然后,构建一个卷积神经网络模型,用来学习用户和兴趣点在各自属性（如用户年龄与兴趣点位置之间）上的显式关联关系;同时,构建另外一个神经网络模型,模拟机器学习中的矩阵分解方法,根据用户的签到行为,深入挖掘用户与兴趣点之间的隐式关联关系。最后,将用户与兴趣点之间的显式和隐式关联关系进行集成,综合表征用户?兴趣点之间的耦合关系,然后将学习到的用户?兴趣点耦合关系输入到一个全连接网络中进行兴趣点推荐。本文所提出的模型在Yelp数据集上进行了评估,实验结果表明该模型在兴趣点推荐方面有较高的推荐准确性。     

基于图卷积与外积的协同过滤推荐模型

推荐系统帮助用户主动找到满足其偏好的个性化物品并推荐给用户.协同过滤算法是推荐系统中较为经典的算法,但是其会受到数据冷启动和稀疏性的限制,具有可解释性差和模型泛化能力差等缺点.针对其缺点进行研究,通过将原始的评分矩阵以用户—项目二部图的形式作为输入,将图卷积神经网络设计为一种图自编码器的变体,通过迭代的聚合邻居节点信息得到用户和项目的潜在向量表示,并在其基础上结合卷积神经网络,提出了一种基于卷积矩阵分解的推荐算法,提升了模型的可解释性和泛化能力,同时融合辅助信息也解决了数据的稀疏性问题,并使推荐的性能分别得到了1.4％和1.7％的提升.为今后在基于图神经网络的推荐方向上提供了一种新的思路.     

基于混合自编码器的协同过滤推荐算法优化

协同过滤算法已广泛应用在推荐系统中,在实现新异性推荐功能中效果显著,但仍存在数据稀疏、扩展性差、冷启动等问题,需要新的设计思路和技术方法进行优化.近几年,深度学习在图像处理、目标识别、自然语言处理等领域均取得突出成果,将深度神经网络模型与推荐算法结合,为构建新型推荐系统带来新的契机.本文提出一种新式混合神经网络模型,该模型由栈式降噪自编码器和深度神经网络构成,学习得到用户和项目的潜在特征向量以及用户-项目之间的交互行为模型,有效解决数据稀疏问题从而提高系统推荐质量.该推荐算法模型通过MovieLens电影评分数据集测试,实验结果与SVD、PMF等传统推荐算法和经典自编码器模型算法作对比,其推荐质量得到显著提升.     

Chinese medical question answer selection via hybrid models based on CNN and GRU

Question answer selection in the Chinese medical field is very challenging since it requires effective text representations to capture the complex semantic relationships between Chinese questions and answers. Recent approaches on deep learning, e.g., CNN and RNN, have shown their potential in improving the selection quality. However, these existing methods can only capture a part or one-side of semantic relationships while ignoring the other rich and sophisticated ones, leading to limited performance improvement. In this paper, a series of neural network models are proposed to address Chinese medical question answer selection issue. In order to model the complex relationships between questions and answers, we develop both single and hybrid models with CNN and GRU to combine the merits of different neural network architectures. This is different from existing works that can onpy capture partial relationships by utilizing a single network structure. Extensive experimental results on cMedQA dataset demonstrate that the proposed hybrid models, especially BiGRU-CNN, significantly outperform the state-of-the-art methods. The source codes of our models are available in the GitHub (https://github.com/zhangyuteng/MedicalQA-CNN-BiGRU).

     

Indoor scene understanding via RGB-D image segmentation employing depth-based CNN and CRFs

With the availability of low-cost depth-visual sensing devices, such as Microsoft Kinect, we are experiencing a growing interest in indoor environment understanding, at the core of which is semantic segmentation in RGB-D image. The latest research shows that the convolutional neural network (CNN) still dominates the image semantic segmentation field. However, down-sampling operated during the training process of CNNs leads to unclear segmentation boundaries and poor classification accuracy. To address this problem, in this paper, we propose a novel end-to-end deep architecture, termed FuseCRFNet, which seamlessly incorporates a fully-connected Conditional Random Fields (CRFs) model into a depth-based CNN framework. The proposed segmentation method uses the properties of pixel-to-pixel relationships to increase the accuracy of image semantic segmentation. More importantly, we formulate the CRF as one of the layers in FuseCRFNet to refine the coarse segmentation in the forward propagation, in meanwhile, it passes back the errors to facilitate the training. The performance of our FuseCRFNet is evaluated by experimenting with SUN RGB-D dataset, and the results show that the proposed algorithm is superior to existing semantic segmentation algorithms with an improvement in accuracy of at least 2%, further verifying the effectiveness of the algorithm.

     

AutoTrustRec: Recommender System with Social Trust and Deep Learning using AutoEncoder

Deep learning is the most active research topic amongst data scientists and analysts these days. It is because deep learning has provided very high accuracy in various domains such as speech recognition, image processing and natural language processing. Researchers are actively working to deploy deep learning on information retrieval. Due to large-scale data generated by social media and sensor networks, it is quite difficult to train unstructured and highly complex data. Recommender system is intelligent information filtering technique which assists the user to find topic of interest within complex overloaded information. In this paper, our motive is to improve recommendation accuracy for large-scale heterogeneous complex data by integrating deep learning architecture. In our proposed approach ratings, direct and indirect trust values are fed in neural network using shared layer in autoencoder. Comprehensive experiment analysis on three public datasets proves that RMSE and MAE are improved significantly by using our proposed approach.

     

Position Encoding Based Convolutional Neural Networks for Machine Remaining Useful Life Prediction

Accurate remaining useful life (RUL) prediction is important in industrial systems. It prevents machines from working under failure conditions, and ensures that the industrial system works reliably and efficiently. Recently, many deep learning based methods have been proposed to predict RUL. Among these methods, recurrent neural network (RNN) based approaches show a strong capability of capturing sequential information. This allows RNN based methods to perform better than convolutional neural network (CNN) based approaches on the RUL prediction task. In this paper, we question this common paradigm and argue that existing CNN based approaches are not designed according to the classic principles of CNN, which reduces their performances. Additionally, the capacity of capturing sequential information is highly affected by the receptive field of CNN, which is neglected by existing CNN based methods. To solve these problems, we propose a series of new CNNs, which show competitive results to RNN based methods. Compared with RNN, CNN processes the input signals in parallel so that the temporal sequence is not easily determined. To alleviate this issue, a position encoding scheme is developed to enhance the sequential information encoded by a CNN. Hence, our proposed position encoding based CNN called PE-Net is further improved and even performs better than RNN based methods. Extensive experiments are conducted on the C-MAPSS dataset, where our PE-Net shows state-of-the-art performance.      

Deep neural network and 3D model for face recognition with multiple disturbing environments

This paper presents the proposed bird search-based shuffled shepherd optimization algorithm (BSSSO) for face recognition. Initially, the input image undergoes a noise removal phase to eliminate noise in order to make them suitable for subsequent processing. The noise removal is performed using the type II fuzzy system and cuckoo search optimization algorithm (T2FCS), which detects noisy pixels from the image for improved processing. After the noise removal phase, the feature extraction is carried out using the convolution neural network (CNN) model and landmark enabled 3D morphable model (L3DMM). The obtained features are subjected to deep CNN for face recognition. The training of deep CNN is performed using the bird search-based shuffled shepherd optimization algorithm (BSSSO). Here, the proposed BSSSO is designed by combining the shuffled shepherd optimization algorithm (SSOA) and bird swarm algorithm (BSA) for inheriting the merits of both optimizations towards effective training of deep CNN. The proposed method obtained higher accuracy of 0.8935 and minimum FAR and FRR of 0.2190 and 0.2021 using LFW database with respect to training data.

     

Optimal Artificial Intelligence Based Automated Skin Lesion Detection and Classification Model

Skin lesions have become a critical illness worldwide, and the earlier identification of skin lesions using dermoscopic images can raise the survival rate. Classification of the skin lesion from those dermoscopic images will be a tedious task. The accuracy of the classification of skin lesions is improved by the use of deep learning models. Recently, convolutional neural networks (CNN) have been established in this domain, and their techniques are extremely established for feature extraction, leading to enhanced classification. With this motivation, this study focuses on the design of artificial intelligence (AI) based solutions, particularly deep learning (DL) algorithms, to distinguish malignant skin lesions from benign lesions in dermoscopic images. This study presents an automated skin lesion detection and classification technique utilizing optimized stacked sparse autoencoder (OSSAE) based feature extractor with backpropagation neural network (BPNN), named the OSSAE-BPNN technique. The proposed technique contains a multi-level thresholding based segmentation technique for detecting the affected lesion region. In addition, the OSSAE based feature extractor and BPNN based classifier are employed for skin lesion diagnosis. Moreover, the parameter tuning of the SSAE model is carried out by the use of sea gull optimization (SGO) algorithm. To showcase the enhanced outcomes of the OSSAE-BPNN model, a comprehensive experimental analysis is performed on the benchmark dataset. The experimental findings demonstrated that the OSSAE-BPNN approach outperformed other current strategies in terms of several assessment metrics.     

A correlative denoising autoencoder to model social influence for top-N recommender system

In recent years,there are numerous works been proposed to leverage the techniques of deep learning to improve social-aware recommendation performance.In most cases,it requires a larger number of data to train a robust deep learning model,which contains a lot of parameters to fit training data.However,both data of user ratings and social networks are facing critical sparse problem,which makes it not easy to train a robust deep neural network model.Towards this problem,we propose a novel correlative denoising autoencoder(CoDAE)method by taking correlations between users with multiple roles into account to learn robust representations from sparse inputs of ratings and social networks for recommendation.We develop the CoDAE model by utilizing three separated autoencoders to learn user features with roles of rater,truster and trustee,respectively.Especially,on account of that each input unit of user vectors with roles of truster and trustee is corresponding to a particular user,we propose to utilize shared parameters to learn common information of the units that corresponding to same users.Moreover,we propose a related regularization term to learn correlations between user features that learnt by the three subnetworks of CoDAE model.We further conduct a series of experiments to evaluate the proposed method on two public datasets for Top-N recommendation task.The experimental results demonstrate that the proposed model outperforms state-of-the-art algorithms on rank-sensitive metrics of MAP and NDCG.     

Enhanced near-infrared periocular recognition through collaborative rendering of hand crafted and deep features

Periocular recognition leverage from larger feature region and lesser user cooperation, when compared against the traditional iris recognition. Moreover, in the current scenario of Covid-19, where majority of people cover their faces with masks, potential of recognizing faces gets reduced by a large extent, calling for wide applicability of periocular recognition. In view of these facts, this paper targets towards enhanced representation of near-infrared periocular images, by combined use of hand-crafted and deep features. The hand-crafted features are extracted through partitioning of periocular image followed by obtaining the local statistical properties pertaining to each partition. Whereas, deep features are extracted through the popular convolutional neural network (CNN) ResNet-101 model. The extensive set of experiments performed with a benchmark periocular database validates the promising performance of the proposed method. Additionally, investigation of cross-spectral matching framework and comparison with state-of-the-art, reveal that combination of both types of features employed could prove to be extremely effective.

     

Improving the readability and saliency of abstractive text summarization using combination of deep neural networks equipped with auxiliary attention mechanism

Rapid and exponential development of textual data in recent years has yielded to the need for automatic text summarization models which aim to automatically condense a piece of text into a shorter version. Although various unsupervised and machine learning-based approaches have been introduced for text summarization during the last decades, the emergence of deep learning has made remarkable progress in this field. However, deep learning-based text summarization models are still in their early steps of development and their potential has yet to be fully explored. Accordingly, a novel abstractive summarization model is proposed in this paper which utilized the combination of convolutional neural network and long short-term memory integrated with auxiliary attention in its encoder to increase the saliency and coherency of generated summaries. The proposed model was validated on CNN\Daily Mail and DUC-2004 datasets and empirical results indicated that not only the proposed model outperformed existing models in terms of ROUGE metric but also its generated summaries had higher saliency and readability compared to the baseline model according to human evaluation.

     

Learn class hierarchy using convolutional neural networks

A large amount of research on Convolutional Neural Networks (CNN) has focused on flat Classification in the multi-class domain. In the real world, many problems are naturally expressed as hierarchical classification problems, in which the classes to be predicted are organized in a hierarchy of classes. In this paper, we propose a new architecture for hierarchical classification, introducing a stack of deep linear layers using cross-entropy loss functions combined to a center loss function. The proposed architecture can extend any neural network model and simultaneously optimizes loss functions to discover local hierarchical class relationships and a loss function to discover global information from the whole class hierarchy while penalizing class hierarchy violations. We experimentally show that our hierarchical classifier presents advantages to the traditional classification approaches finding application in computer vision tasks. The same approach can also be applied to some CNN for text classification.

     

Development of acoustic source localization with adaptive neural network using distance mating-based red deer algorithm

Multichannel, audio processing approaches are widely examined in human–computer interaction, autonomous robots, audio surveillance, and teleconferencing systems. The numerous applications are linked to the speech technology and acoustic analysis area. Much attention is received to the active speakers and spatial localization of acoustic sources on the acoustic sensor arrays. Baseline approaches provide negotiable performance in a real-world comprised of far-field/near-field monitoring, reverberant and noisy environments, and also the outdoor/indoor scenarios. A practical system to detect defects in complex structures is the time difference mapping (TDM) technique. The significant scope of the research is to search the location using the minimum distance point in the time difference database to be apart from the verification point. In the case of the improved “time difference mapping (I-TDM)” technique and traditional “time difference mapping (T-TDM)” technique, the denser grids and vast database permit increased accuracy. In the database, if the location points are not present, then the accurate localization of the I-TDM and T-TDM techniques is damaged. Hence, to handle these problems, this article plans to develop acoustic source localization according to the deep learning strategy. The audio dataset is gathered from the benchmark source called the SSLR dataset and is initially subjected to preprocessing, which involves artifact removal and smoothing for effective processing. Further, the adaptive convolutional neural network (CNN)-based feature set creation is performed. Here, the adaptive CNN is accomplished by the improved optimization algorithm called distance mating-based red deer algorithm (DM-RDA). With this trained feature set, the acoustic source localization is done by the weight updated deep neural network, in which the same DM-RDA is used for optimizing the training weight. The simulation outcome proves that the designed model produced enhanced performance compared to other traditional source localization estimators.     

Enhancing train position perception through AI-drivenmulti-source information fusion

This paper addresses the challenge of accurately and timely determining the position of a train, with specific consideration given to the integration of the global navigation satellite system (GNSS) and inertial navigation system (INS). To overcome the increasing errors in the INS during interruptions in GNSS signals, as well as the uncertainty associated with process and measurement noise, a deep learning-based method for train positioning is proposed. This method combines convolutional neural networks (CNN), long short-term memory (LSTM), and the invariant extended Kalman filter (IEKF) to enhance the perception of train positions. It effectively handles GNSS signal interruptions and mitigates the impact of noise. Experimental evaluation and comparisons with existing approaches are provided to illustrate the effectiveness and robustness of the proposed method.     

基于小样本学习的SAR图像识别

深度学习已成为图像识别领域的一个研究热点。与传统图像识别方法不同,深度学习从大量数据中自动学习特征,并且具有强大的自学习能力和高效的特征表达能力。但在小样本条件下,传统的深度学习方法如卷积神经网络难以学习到有效的特征,造成图像识别的准确率较低。因此,提出一种新的小样本条件下的图像识别算法用于解决SAR图像的分类识别。该算法以卷积神经网络为基础,结合自编码器,形成深度卷积自编码网络结构。首先对图像进行预处理,使用2D Gabor滤波增强图像,在此基础上对模型进行训练,最后构建图像分类模型。该算法设计的网络结构能自动学习并提取小样本图像中的有效特征,进而提高识别准确率。在MSTAR数据集的10类目标分类中,选择训练集数据中10%的样本作为新的训练数据,其余数据为验证数据,并且,测试数据在卷积神经网络中的识别准确率为76.38%,而在提出的卷积自编码结构中的识别准确率达到了88.09%。实验结果表明,提出的算法在小样本图像识别中比卷积神经网络模型更加有效。     

Accessible wayfinding and navigation: a systematic mapping study

Urban environments, university campuses, and public and private buildings often present architectural barriers that prevent people with disabilities and special needs to move freely and independently. This paper presents a systematic mapping study of the scientific literature proposing devices, and software applications aimed at fostering accessible wayfinding and navigation in indoor and outdoor environments. We selected 111 out of 806 papers published in the period 2009–2020, and we analyzed them according to different dimensions: at first, we surveyed which solutions have been proposed to address the considered problem; then, we analyzed the selected papers according to five dimensions: context of use, target users, hardware/software technologies, type of data sources, and user role in system design and evaluation. Our findings highlight trends and gaps related to these dimensions. The paper finally presents a reflection on challenges and open issues that must be taken into consideration for the design of future accessible places and of related technologies and applications aimed at facilitating wayfinding and navigation.

     

基于深度学习的GPS轨迹去匿名研究

移动互联网和LBS技术的高速发展使得位置服务提供商可以轻松收集到大量用户位置轨迹数据,近期研究表明,深度学习方法能够从轨迹数据集中提取出用户身份标识等隐私信息.然而现有工作主要针对社交网络采集的签到点轨迹,针对GPS轨迹的去匿名研究则较为缺乏.因此,对基于深度学习的GPS轨迹去匿名技术开展研究.首先提出一种GPS轨迹数...