Multi-scale context-aware network for continuous sign language recognition

The hands and face are the most important parts for expressing sign language morphemes in sign language videos.However,we find that existing Continuous Sign Language Recognition(CSLR)methods lack the mining of hand and face information in visual backbones or use expensive and time-consuming external extractors to explore this information.In addition,the signs have different lengths,whereas previous CSLR methods typically use a fixed-length window to segment the video to capture sequential features and then perform global temporal modeling,which disturbs the perception of complete signs.In this study,we propose a Multi-Scale Context-Aware network(MSCA-Net)to solve the aforementioned problems.Our MSCA-Net contains two main modules:(1)Multi-Scale Motion Attention(MSMA),which uses the differences among frames to perceive information of the hands and face in multiple spatial scales,replacing the heavy feature extractors;and(2)Multi-Scale Temporal Modeling(MSTM),which explores crucial temporal information in the sign language video from different temporal scales.We conduct extensive experiments using three widely used sign language datasets,i.e.,RWTH-PHOENIX-Weather-2014,RWTH-PHOENIX-Weather-2014T,and CSL-Daily.The proposed MSCA-Net achieve state-of-the-art performance,demonstrating the effectiveness of our approach.

Smaller & Smarter: Score-Driven Network Chaining of Smaller Language Models

With the continuous evolution and expanding applications of Large Language Models (LLMs), there has been a noticeable surge in the size of the emerging models. It is not solely the growth in model size, primarily measured by the number of parameters, but also the subsequent escalation in computational demands, hardware and software prerequisites for training, all culminating in a substantial financial investment as well. In this paper, we present novel techniques like supervision, parallelization, and scoring functions to get better results out of chains of smaller language models, rather than relying solely on scaling up model size. Firstly, we propose an approach to quantify the performance of a Smaller Language Models (SLM) by introducing a corresponding supervisor model that incrementally corrects the encountered errors. Secondly, we propose an approach to utilize two smaller language models (in a network) performing the same task and retrieving the best relevant output from the two, ensuring peak performance for a specific task. Experimental evaluations establish the quantitative accuracy improvements on financial reasoning and arithmetic calculation tasks from utilizing techniques like supervisor models (in a network of model scenario), threshold scoring and parallel processing over a baseline study.

On the Application of Task-based Language Teaching Approach to English Phonetic Teaching

Task-based language teaching approach(TBLTA), which lays stress on "learning by doing", gained increasing popularity in English teaching in recent years. The design of phonetic teaching calls for more emphasis from English educators since it is one of the basic rounds of English teaching. This paper made a trial on the utilization of TBLTA in the English phonetic teaching context and designed a TBLTA model for English phonetic teaching based on discussions about model and merits of TBLTA.

On the clash between the traditional culture and the ＂fast food＂ culture from a perspective of internet language

With the development of the Internet, the Internet language-a new social variant of language comes into being. Owing to its distinctive features, the Internet language draws extensive attention and evokes dispute. Should we make laws to restrict its development or be magnanimous and tolerant？ The present authors think that the dispute reflects the clash between the traditional culture and the ＂fast food＂ culture. The two cultures would co-exist harmoniously and co-develop healthily only if we cherish the law in the development of language and grasp the indirect correlation between them.

An Experimental Study of Network-based Language Learning in Less-developed Areas

The advent of the Age of Information brings about bright prospects to Network-based Language Learning(NBLL).The thesis adopts the Engagement Theory as guided principles.The purpose is to use the novel NBLL model effectively with the help of modern technology especially in less-developed areas.This thesis focuses on network-based experimental study.The research shows that the students under NBLL environment have cultivated the capabilities in information collection,computer operation,and information evaluation,as well as the abilities in problem solving,reasoning with criticism,and cooperating with others.

Internet English and Its Impact on Language Teaching

With the development of economic globalization,English has been accorded paramount importance.Today,English is the most widely used language in the world and is one of the most important information vehicles.Also the Internet has unveiled a new era of mass communication in human history and its development has great effect on English.This article presents the effect that Internet brought on English development from different aspects and analyzes the reasons why these changes have taken place.Also this essay provides the benefits and challenges of Internet for second and foreign language acquisition.

Evidence of cortical reorganization of language networks after stroke with subacute Broca's aphasia:a blood oxygenation level dependent-functional magnetic resonance imaging study

Aphasia is an acquired language disorder that is a common consequence of stroke.The pathogenesis of the disease is not fully understood,and as a result,current treatment options are not satisfactory.Here,we used blood oxygenation level-dependent functional magnetic resonance imaging to evaluate the activation of bilateral cortices in patients with Broca＇s aphasia 1 to 3 months after stroke.Our results showed that language expression was associated with multiple brain regions in which the right hemisphere participated in the generation of language.The activation areas in the left hemisphere of aphasia patients were significantly smaller compared with those in healthy adults.The activation frequency,volumes,and intensity in the regions related to language,such as the left inferior frontal gyrus（Broca＇s area）,the left superior temporal gyrus,and the right inferior frontal gyrus（the mirror region of Broca＇s area）,were lower in patients compared with healthy adults.In contrast,activation in the right superior temporal gyrus,the bilateral superior parietal lobule,and the left inferior temporal gyrus was stronger in patients compared with healthy controls.These results suggest that the right inferior frontal gyrus plays a role in the recovery of language function in the subacute stage of stroke-related aphasia by increasing the engagement of related brain areas.

Evaluating the Efficiency of CBAM-Resnet Using Malaysian Sign Language

The deaf-mutes population is constantly feeling helpless when others do not understand them and vice versa.To fill this gap,this study implements a CNN-based neural network,Convolutional Based Attention Module(CBAM),to recognise Malaysian Sign Language(MSL)in videos recognition.This study has created 2071 videos for 19 dynamic signs.Two different experiments were conducted for dynamic signs,using CBAM-3DResNet implementing‘Within Blocks’and‘Before Classifier’methods.Various metrics such as the accuracy,loss,precision,recall,F1-score,confusion matrix,and training time were recorded to evaluate the models’efficiency.Results showed that CBAM-ResNet models had good performances in videos recognition tasks,with recognition rates of over 90%with little variations.CBAMResNet‘Before Classifier’is more efficient than‘Within Blocks’models of CBAM-ResNet.All experiment results indicated the CBAM-ResNet‘Before Classifier’efficiency in recognising Malaysian Sign Language and its worth of future research.

Politics in Language: on George Orwell's View of Language in Nineteen Eighty-Four

Language stood in the foreground of George Orwell's social and political thinking. Language is not only a vehicle for transmitting ideas, but also a product originating from social and political interactions. This paper examines the language, particularly the abuse of language, in the context of politics presented in Nineteen Eighty-Four, so as to find out George Orwell's view of language: language as a manipulated tool to control people's minds.

Computer Assisted Language Learning Social Networks: What Are They Talking about?

There are many reasons that motivate people to build online communities. The purpose of this study was to identify the topics that learners discuss when they are part of a computer assisted language learning course in order to answer the question “What are they talking about?”. We have examined an e-community of 618 students who were learning the Modern Greek language online. We analyzed their conversation topics directly from the discussion boards of the web-based course and sorted them into the pre-defined topic categories. The results of the study showed that during the first lessons of the course the students contributed more to social discussions which were unrelated to the course material. The reason of this outcome is that the students want to introduce themselves and meet their peers. As they progressed through the course’s lessons, however, their discussion topics became more course material related. The study ends with implications of the results and future research directions.

Recognition of Arabic Sign Language (ArSL) Using Recurrent Neural Networks

The objective of this research is to introduce the use of different types of neural networks in human hand gesture recognition for static images as well as for dynamic gestures. This work focuses on the ability of neural networks to assist in Arabic Sign Language (ArSL) hand gesture recognition. We have presented the use of feedforward neural networks and recurrent neural networks along with its different architectures;partially and fully recurrent networks. Then we have tested our proposed system;the results of the experiment have showed that the suggested system with the fully recurrent architecture has had a performance with an accuracy rate 95% for static gesture recognition.

Numerical‐discrete‐scheme‐incorporated recurrent neural network for tasks in natural language processing

A variety of neural networks have been presented to deal with issues in deep learning in the last decades.Despite the prominent success achieved by the neural network,it still lacks theoretical guidance to design an efficient neural network model,and verifying the performance of a model needs excessive resources.Previous research studies have demonstrated that many existing models can be regarded as different numerical discretizations of differential equations.This connection sheds light on designing an effective recurrent neural network(RNN)by resorting to numerical analysis.Simple RNN is regarded as a discretisation of the forward Euler scheme.Considering the limited solution accuracy of the forward Euler methods,a Taylor‐type discrete scheme is presented with lower truncation error and a Taylor‐type RNN(T‐RNN)is designed with its guidance.Extensive experiments are conducted to evaluate its performance on statistical language models and emotion analysis tasks.The noticeable gains obtained by T‐RNN present its superiority and the feasibility of designing the neural network model using numerical methods.

Continuous Sign Language Recognition Based on Spatial-Temporal Graph Attention Network

Continuous sign language recognition(CSLR)is challenging due to the complexity of video background,hand gesture variability,and temporal modeling difficulties.This work proposes a CSLR method based on a spatialtemporal graph attention network to focus on essential features of video series.The method considers local details of sign language movements by taking the information on joints and bones as inputs and constructing a spatialtemporal graph to reflect inter-frame relevance and physical connections between nodes.The graph-based multihead attention mechanism is utilized with adjacent matrix calculation for better local-feature exploration,and short-term motion correlation modeling is completed via a temporal convolutional network.We adopted BLSTM to learn the long-termdependence and connectionist temporal classification to align the word-level sequences.The proposed method achieves competitive results regarding word error rates(1.59%)on the Chinese Sign Language dataset and the mean Jaccard Index(65.78%)on the ChaLearn LAP Continuous Gesture Dataset.

An Optimized Convolutional Neural Network with Combination Blocks for Chinese Sign Language Identification

(Aim)Chinese sign language is an essential tool for hearing-impaired to live,learn and communicate in deaf communities.Moreover,Chinese sign language plays a significant role in speech therapy and rehabilitation.Chinese sign language identification can provide convenience for those hearing impaired people and eliminate the communication barrier between the deaf community and the rest of society.Similar to the research of many biomedical image processing(such as automatic chest radiograph processing,diagnosis of chest radiological images,etc.),with the rapid development of artificial intelligence,especially deep learning technologies and algorithms,sign language image recognition ushered in the spring.This study aims to propose a novel sign language image recognition method based on an optimized convolutional neural network.(Method)Three different combinations of blocks:Conv-BN-ReLU-Pooling,Conv-BN-ReLU,Conv-BN-ReLU-BN were employed,including some advanced technologies such as batch normalization,dropout,and Leaky ReLU.We proposed an optimized convolutional neural network to identify 1320 sign language images,which was called as CNN-CB method.Totally ten runs were implemented with the hold-out randomly set for each run.(Results)The results indicate that our CNN-CB method gained an overall accuracy of 94.88±0.99%.(Conclusion)Our CNN-CB method is superior to thirteen state-of-the-art methods:eight traditional machine learning approaches and five modern convolutional neural network approaches.

Automatic Mexican Sign Language Recognition Using Normalized Moments and Artificial Neural Networks

This document presents a computer vision system for the automatic recognition of Mexican Sign Language (MSL), based on normalized moments as invariant (to translation and scale transforms) descriptors, using artificial neural networks as pattern recognition model. An experimental feature selection was performed to reduce computational costs due to this work focusing on automatic recognition. The computer vision system includes four LED-reflectors of 700 lumens each in order to improve image acquisition quality;this illumination system allows reducing shadows in each sign of the MSL. MSL contains 27 signs in total but 6 of them are expressed with movement;this paper presents a framework for the automatic recognition of 21 static signs of MSL. The proposed system achieved 93% of recognition rate.

NNL:a domain-specific language for neural networks

Recent years,neural networks(NNs)have received increasing attention from both academia and industry.So far significant diversity among existing NNs as well as their hardware platforms makes NN programming a daunting task.In this paper,a domain-specific language(DSL)for NNs,neural network language(NNL)is proposed to deliver productivity of NN programming and portable performance of NN execution on different hardware platforms.The productivity and flexibility of NN programming are enabled by abstracting NNs as a directed graph of blocks.The language describes 4 representative and widely used NNs and runs them on 3 different hardware platforms(CPU,GPU and NN accelerator).Experimental results show that NNs written with the proposed language are,on average,14.5%better than the baseline implementations across these 3 platforms.Moreover,compared with the Caffe framework that specifically targets the GPU platform,the code can achieve similar performance.

“NET三维联动”产教融合外语人才培养中的课程思政实施路径

“NET三维联动”产教融合应用型外语人才培养模式,可通过“课内与课外、学校与企业、校内导师与企业导师”三维联动,将课程思政教育和外语专业实践能力培养从课堂延展到课外活动、社会实践、企业服务,建立网络资源库(Network),向合作企业开展志愿外语翻译服务(Enterprise),校内外双导师(Tutor)共同培养应用型外语人才。

STUDY ON NATURAL LANGUAGE INTERFACE OF NETWORK FAULT DIAGNOSIS EXPERT SYSTEM

The expert system is an important field of the artificial intelligence. The traditional interface of the expert system is the command, menu and window at present. It limits the application of the expert system and embarrasses the enthusiasm of using expert system. Combining with the study on the expert system of network fault diagnosis, the natural language interface of the expert system has been discussed in this article. This interface can understand and generate Chinese sentences. Using this interface, the user and field experts can use the expert system to diagnose the fault of network conveniently. In the article, first, the extended production rule has been proposed. Then the methods of Chinese sentence generation from conceptual graphs and the model of expert system are introduced in detail. Using this model, the network fault diagnosis expert system and its natural language interface have been developed with Prolog.

Interrelation of Languages of Colored Petri Nets and Some Traditional Languages

The article studies the interrelation of Languages of Colored Petri Nets and Traditional formal languages. The author constructed the graph of Colored Petri Net, which generates L* Context-free language. This language may not be modeled using standard Petri Nets [1]. The Venn graph and diagram that the author modified [1], show the interrelation between languages of Colored Petri Nets and some Traditional languages. Thus the class of languages of Colored Petri Nets is supposed to include an entire class of Context-free languages.

Greek Political Language during the Economic Crisis—A Network Analytic Approach

In this paper, we seek to analyze pre-electoral political language in Greece with the use of Social Network Analysis. For this analysis, we collected data from the pre-elections speeches of five political leaders from the 20th of September 2015 Greek general elections. We proceed to form, analyze and compare networks of words with an emphasis on financial vocabulary. Findings can provide interesting insights into how political leaders structure their speeches, evaluate important issues and use economic terms and political rhetoric, while different structural patterns can reveal the differences between political parties. Finally, we check whether the overall networks follow the general rules of real-life networks by belonging to the small-world or scale-free categories.