Textual Emotion Analysis (TEA) aims to extract and analyze user emotional states in texts. Various Deep Learning (DL) methods have developed rapidly, they proven be successful many fields such as audio, image, natural language processing. This trend has drawn increasing researchers away from traditional machine learning DL for their scientific research. In this paper, we provide an overview of TEA based on methods. After introducing a background emotion analysis that includes defining...

Deep convolutional neural networks (CNNs) have gained great success in various computer vision applications. State-of-the-art CNN models for large-scale applications are computation intensive and memory expensive and, hence, mainly processed on high-performance processors like server CPUs GPUs. However, there is an increasing demand of high-accuracy or real-time object detection tasks clusters embedded systems, which requires energy-efficient accelerators because the green requirement...

The goal of cross-domain sentiment classification is to utilise useful information in the source domain help classify polarity target domain, which has a large number unlabelled data. Most existing methods focus on extracting invariant features between two domains. But they cannot make better use data domain. To solve this problem, we present deep transfer learning mechanism (DTLM) for fine-grained classification. DTLM provides across domains by incorporating BERT(Bidirextional Encoder...

Deep convolutional neural networks (CNN) is highly efficient in image recognition tasks such as MNIST digit recognition. Accelerators based on FPGA platform are proposed since general purpose processor disappointing terms of performance when dealing with tasks. Recently, an optimized FPGA-based accelerator design (work 1) has been claiming best compared existing implementations. But the author acknowledged, could be better if fixed point presentation and computation elements had used....

Social networks are critical in terms of information or malware propagation. However, how to contain the spreading social is still an open and challenging issue. In this paper, we propose a novel defending method through big data based influence modeling. We first establish interaction graph on sets studied object. Based graph, able measure direct individuals by computing each node's strength, which includes degree node total number messages sent user her friends. Then, design algorithm...

The Web of Things (WoT) extends the concept ''Internet (IoT)'' in that smart devices physical world can be interacted with or integrated via popular web technologies (e.g., HTML, HTTP, and API).With WoT, use APIs to make their data functionalities accessible by software.With popularization 2.0 Mashup applications, creating applications for IoT (or WoT) combining different APIs, also has aroused increasing interests.This paper proposes an approach mining collaboration patterns between aid...

A method of region-based image segmentation with mean-shift clustering algorithm is introduced. This first extracts color, texture, and location features from each pixel to form feature vector by selecting suitable color space. Then, these vectors are the window parameter r decided proposed optimal amount, so numbers centers clusters also selected, grouped labeled. Finally, regions same label segmented again according neighbor connection theory for pixels a lot which describe provided....

Negative emotion classification refers to the automatic of negative texts in social networks. Most existing methods are based on deep learning models, facing challenges such as complex structures and too many hyperparameters. To meet these challenges, this paper, we propose a method for utilizing Robustly Optimized BERT Pretraining Approach (RoBERTa) <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$p$</tex> -norm Broad Learning ( -BL)....

The task of retrieving audio content relevant to lyric queries and vice versa plays a critical role in music-oriented applications. In this process, robust feature representations have be learned for two modalities. Furthermore, interactions between different modalities should properly captured at fine-grained level. Existing approaches can effectively extract modal perform through alignment. However, these model lyrics coarse-grained manner. Especially the input features enhanced produced...

A proposed KFCM-based fuzzy classifier was introduced. As for the process of constructing such classifier, firstly, original sample space is mapped into a high dimensional feature by selecting appropriate kernel function. Then in space, training samples each class are divided some clusters KFCM algorithm. The optimal cluster number every selected our method. For created cluster, rule defined with ellipsoidal regions. Finally, classification rules adjusted GAs. accuracy constructed methods...

The spread of fake news has caused severe damage to people's lives and society nowadays. Social media is inundated with from multiple domains. Previously, the methods used detect have tended be limited single domains performed inadequacy in other or Therefore, detection multi-domain garnered significant interest. However, approaches rely more on sufficient training samples. In real world, low-resource problem become a challenge that restricts multi-domain. case, prompt learning advantages...

The Chinese short text sentiment classification method based on word embedding and convolutional neural network has already achieved effective results. recent mainly adopts single channel CNN. In order to enhance performance, this study proposes a Dual-channel word2vec semantic feature topic model (LDA)feature of words were used as input CNN extract sequence sentimental features which, subsequently applied classification. experimental results show that the performance is better than those...

Similar Case Matching (SCM) is a vital component of legal retrieval and plays crucial role in intelligent systems. This task endeavors to ascertain the similarity between two cases. Recently, SCM methods have utilized pre-trained language model (PLM) encode case representations, following traditional pre-train fine-tune approach. How- ever, due inconsistency objective PLM matching, we argue that this approach does not effectively leverage rich semantic linguistic knowledge accumulated during...

With the rapid development of web technology, Social Networks (SNs) have become one most popular platforms for users to exchange views and express their emotions. More more people are used commenting on a certain hot spot in SNs, resulting large amount texts containing Textual Emotion Cause Extraction (TECE) aims automatically extract causes emotion texts, which is an important research issue natural language processing. It different from previous tasks recognition classification. In...

The prevalence of fake news online has become a significant societal concern. To combat this, multimodal detection techniques based on images and text have shown promise. Yet, these methods struggle to analyze complex relationships within between modalities due the diverse discriminative elements in content. In addition, research multi-class remains insufficient. address above challenges, this paper, we propose novel model, GS 2 F, leveraging g raph s tructure uided emantic f usion....

In this paper, a novel fuzzy classifier ensemble system is proposed. This can reduce subjective factor in building classifier, and improve the classification recognition rate stability. Three proposed approaches are introduced, namely, approach of measuring generalization difference(GD) sets to select individual classifiers, determining classifier's reliability by membership matrix, ensemble. The evaluated with standard data sets. comparison experiments existed systems. experiment results...