FLOCKING OF A THERMODYNAMIC CUCKER-SMALE MODEL WITH LOCAL VELOCITY INTERACTIONS

In this paper, we study the flocking behavior of a thermodynamic Cucker–Smale model with local velocity interactions. Using the spectral gap of a connected stochastic matrix, together with an elaborate estimate on perturbations of a linearized system, we provide a sufficient framework in terms of initial data and model parameters to guarantee flocking. Moreover, it is shown that the system achieves a consensus at an exponential rate.

Lateral interaction by Laplacian‐based graph smoothing for deep neural networks

Lateral interaction in the biological brain is a key mechanism that underlies higher cognitive functions.Linear self‐organising map(SOM)introduces lateral interaction in a general form in which signals of any modality can be used.Some approaches directly incorporate SOM learning rules into neural networks,but incur complex operations and poor extendibility.The efficient way to implement lateral interaction in deep neural networks is not well established.The use of Laplacian Matrix‐based Smoothing(LS)regularisation is proposed for implementing lateral interaction in a concise form.The authors’derivation and experiments show that lateral interaction implemented by SOM model is a special case of LS‐regulated k‐means,and they both show the topology‐preserving capability.The authors also verify that LS‐regularisation can be used in conjunction with the end‐to‐end training paradigm in deep auto‐encoders.Additionally,the benefits of LS‐regularisation in relaxing the requirement of parameter initialisation in various models and improving the classification performance of prototype classifiers are evaluated.Furthermore,the topologically ordered structure introduced by LS‐regularisation in feature extractor can improve the generalisation performance on classification tasks.Overall,LS‐regularisation is an effective and efficient way to implement lateral interaction and can be easily extended to different models.

Graph-based method for human-object interactions detection

Human-object interaction(HOIs)detection is a new branch of visual relationship detection,which plays an important role in the field of image understanding.Because of the complexity and diversity of image content,the detection of HOIs is still an onerous challenge.Unlike most of the current works for HOIs detection which only rely on the pairwise information of a human and an object,we propose a graph-based HOIs detection method that models context and global structure information.Firstly,to better utilize the relations between humans and objects,the detected humans and objects are regarded as nodes to construct a fully connected undirected graph,and the graph is pruned to obtain an HOI graph that only preserving the edges connecting human and object nodes.Then,in order to obtain more robust features of human and object nodes,two different attention-based feature extraction networks are proposed,which model global and local contexts respectively.Finally,the graph attention network is introduced to pass messages between different nodes in the HOI graph iteratively,and detect the potential HOIs.Experiments on V-COCO and HICO-DET datasets verify the effectiveness of the proposed method,and show that it is superior to many existing methods.

ST-SIGMA:Spatio-temporal semantics and interaction graph aggregation for multi-agent perception and trajectory forecasting

Scene perception and trajectory forecasting are two fundamental challenges that are crucial to a safe and reliable autonomous driving(AD)system.However,most proposed methods aim at addressing one of the two challenges mentioned above with a single model.To tackle this dilemma,this paper proposes spatio-temporal semantics and interaction graph aggregation for multi-agent perception and trajectory forecasting(STSIGMA),an efficient end-to-end method to jointly and accurately perceive the AD environment and forecast the trajectories of the surrounding traffic agents within a unified framework.ST-SIGMA adopts a trident encoder-decoder architecture to learn scene semantics and agent interaction information on bird’s-eye view(BEV)maps simultaneously.Specifically,an iterative aggregation network is first employed as the scene semantic encoder(SSE)to learn diverse scene information.To preserve dynamic interactions of traffic agents,ST-SIGMA further exploits a spatio-temporal graph network as the graph interaction encoder.Meanwhile,a simple yet efficient feature fusion method to fuse semantic and interaction features into a unified feature space as the input to a novel hierarchical aggregation decoder for downstream prediction tasks is designed.Extensive experiments on the nuScenes data set have demonstrated that the proposed ST-SIGMA achieves significant improvements compared to the state-of-theart(SOTA)methods in terms of scene perception and trajectory forecasting,respectively.Therefore,the proposed approach outperforms SOTA in terms of model generalisation and robustness and is therefore more feasible for deployment in realworld AD scenarios.

Human-Object Interaction Recognition Based on Modeling Context

This paper proposes a method to recognize human-object interactions by modeling context between human actions and interacted objects.Human-object interaction recognition is a challenging task due to severe occlusion between human and objects during the interacting process.Since that human actions and interacted objects provide strong context information,i.e.some actions are usually related to some specific objects,the accuracy of recognition is significantly improved for both of them.Through the proposed method,both global and local temporal features from skeleton sequences are extracted to model human actions.In the meantime,kernel features are utilized to describe interacted objects.Finally,all possible solutions from actions and objects are optimized by modeling the context between them.The results of experiments demonstrate the effectiveness of our method.

Generation of various multiatom entangled graph states via resonant interactions

In this paper, a scheme for generating various multiatom entangled graph states via resonant interactions is proposed. We investigate the generation of various four-atom graph states first in the ideal case and then in the case in which the cavity decay and atomic spontaneous emission are taken into consideration in the process of interaction. More importantly, we improve the possible distortion of the graph states coming from cavity decay and atomic spontaneous emission by performing appropriate unitary transforms on atoms. The generation of multiatom entangled graph states is very important for constructing quantum one-way computer in a fault-tolerant manner. The resonant interaction time is very short, which is important in the sense of decoherence. Our scheme is easy and feasible within the reach of current experimental technology.

An Intelligent Framework for Recognizing Social Human-Object Interactions

Human object interaction(HOI)recognition plays an important role in the designing of surveillance and monitoring systems for healthcare,sports,education,and public areas.It involves localizing the human and object targets and then identifying the interactions between them.However,it is a challenging task that highly depends on the extraction of robust and distinctive features from the targets and the use of fast and efficient classifiers.Hence,the proposed system offers an automated body-parts-based solution for HOI recognition.This system uses RGB(red,green,blue)images as input and segments the desired parts of the images through a segmentation technique based on the watershed algorithm.Furthermore,a convex hullbased approach for extracting key body parts has also been introduced.After identifying the key body parts,two types of features are extracted.Moreover,the entire feature vector is reduced using a dimensionality reduction technique called t-SNE(t-distributed stochastic neighbor embedding).Finally,a multinomial logistic regression classifier is utilized for identifying class labels.A large publicly available dataset,MPII(Max Planck Institute Informatics)Human Pose,has been used for system evaluation.The results prove the validity of the proposed system as it achieved 87.5%class recognition accuracy.

Gate Feature Interaction Network for Relation Prediction in Knowledge Graph

Recently,many knowledge graph embedding models for knowledge graph completion have been proposed,ranging from the initial translation-based model such as TransE to recent CNN-based models such as ConvE.These models fill in the missing relations between entities by focusing on capturing the representation features to further complete the existing knowledge graph(KG).However,the above KG-based relation prediction research ignores the interaction information among entities in KG.To solve this problem,this work proposes a novel model called Gate Feature Interaction Network(GFINet)with a weighted loss function that takes the benefit of interaction information and deep expressive features together.Specifically,the proposed GFINet consists of a gate convolution block and an interaction attention module,corresponding to catching deep expressive features and interaction information based on these valid features respectively.Our method establishes state-of-the-art experimental results on the standard datasets for knowledge graph completion.In addition,we make ablation experiments to verify the effectiveness of the gate convolution block and the interaction attention module.

Interaction behavior recognition from multiple views

This paper proposed a novel multi-view interactive behavior recognition method based on local self-similarity descriptors and graph shared multi-task learning. First, we proposed the composite interactive feature representation which encodes both the spatial distribution of local motion of interest points and their contexts. Furthermore, local self-similarity descriptor represented by temporal-pyramid bag of words(BOW) was applied to decreasing the influence of observation angle change on recognition and retaining the temporal information. For the purpose of exploring latent correlation between different interactive behaviors from different views and retaining specific information of each behaviors, graph shared multi-task learning was used to learn the corresponding interactive behavior recognition model. Experiment results showed the effectiveness of the proposed method in comparison with other state-of-the-art methods on the public databases CASIA, i3Dpose dataset and self-built database for interactive behavior recognition.

Modeling the Effects of Interactions between Environmental Variables on the State of an Environmental Issue: The Case of the Morelos State in Mexico

An important use of environmental indicators is oriented to know their individual impact on the whole environment quality. Nevertheless, most of the important causes of environment affectations are derived from multiple interactions between indicators which correspond more specifically to the environmental reality. The affectations derived from interactions should be analyzed and interpreted through numerical expressions representing a relevant challenge for developers of environmental indicators. To cope with the analysis and interpretation problem, we propose in this work a methodology in two senses: in a bottom-up sense a directed graph is built representing interactions between environmental indicators as behavioral relations, which exert an effect on the state of an environmental issue of a site over time (10 years);in a top-down sense to assist users in the analysis and interpretation of interactions through a computer interface that provides users with the capacity of knowing how and what relational behaviors between indicators are affecting, the most or the least, the performance of the environmental issue being studied. This methodology was applied to the analysis an interpretation of interactions between environmental variables that affect the state of an environmental quality issue related with the State of Morelos in Mexico. The results showed the adequate expressivity of a directed graph to represent interactions allowed to verify the coherence of the numerical values associated with their behaviors during a period of time and with their effects on the environmental issue under study.

Graph-enhanced neural interactive collaborative filtering

To improve the training efficiency and recommendation accuracy in cold-start interactive recommendation systems,a new graph structure called item similarity graph is proposed on the basis of real data from a public dataset.The proposed graph is built from collaborative interactions and a deep reinforcement learning-based graph-enhanced neural interactive collaborative filtering(GE-ICF)model.The GE-ICF framework is developed with a deep reinforcement learning framework and comprises an embedding propagation layer designed with graph neural networks.Extensive experiments are conducted to investigate the efficiency of the proposed graph structure and the superiority of the proposed GE-ICF framework.Results show that in cold-start interactive recommendation systems,the proposed item similarity graph performs well in data relationship modeling,with the training efficiency showing significant improvement.The proposed GE-ICF framework also demonstrates superiority in decision modeling,thereby increasing the recommendation accuracy remarkably.

基于异质图嵌入和会话交互的课程推荐模型

大规模在线教育平台所形成的网络具有数据量大、实体类型丰富、关系复杂等特性。一方面,在线教育正在被大力普及,而另一方面,在线课程却面临低使用率、低完成度及高辍学率的问题。个性化的课程推荐有利于提高学习者的学习积极性,其中,课程能否顺利合格完成是学习者在选课时所考虑的重要因素。鉴于此,提出一种基于学习完成度预测的个性化课程推荐模型。对学生的课程学习会话图进行建模,根据学生的课程学习顺序以及课程的完成情况,生成学生的学习状态表征;同时考虑在线学习环境因素对课程的影响,构建在线课程学习异质图,采用图神经网络生成异质图中课程节点的嵌入;然后通过交互机制融合学习状态表征和课程嵌入,预测学生下一门将学课程的完成度,根据完成度排序从而实现课程推荐。在CNPC、HMXPC和Scho1at3个大规模在线课程学习数据集上的实验结果表明,该模型能有效提升推荐的准确度,在归一化折损累计增益(NDCG)和平均倒数排名(MRR)2个指标上相较于基线模型最优结果均有显著提升,评估指标K值取5时,其NDCG@5指标在3个数据集上分别提升21.08%、17.73%和5.41%,MRR@5指标在3个数据集上分别提升25.66%、31.59%和26.96%。

基于文献计量分析方法的海气相互作用领域研究态势分析

海洋和大气是地球气候系统关键组成部分,其相互作用对全球气候及人类生活产生深远影响。为了分析海气相互作用的研究现状与未来的发展趋势。采用文献计量分析方法,运用VOSviewer和CiteSpace工具,筛选了中国知网(China National Knowledge Infrastructure,CNKI)和Web of Science核心合集数据库中关于海气相互作用的文献,通过分析关键词共现网络图谱、时间序列图谱、突现网络图谱以及文献的发文国家和机构分布,对1981-2021年中国在该领域的研究进展和未来发展趋势进行了综合评估。同时,本研究还选取了2001-2021年的国际文献,对全球范围内的研究进展和热点问题进行了分析。研究表明,国内外在海气相互作用领域的研究均从宏观到微观角度深入探讨,着重研究了大气和海洋的基本作用机理,并进一步分析了影响地球气候系统的多种因素,目前研究重点转向了极端天气系统的预测及其运行机制。从研究机构和国家分布来看,高校与政府机构是主要的研究主体,国内研究机构之间的合作联系较为紧密。研究关键方向和方法,从宏观层面的基础研究转向更为深入的海气通量和大尺度过程研究。这一转变反映了国际海洋发展战略的调整,同时为未来的研究方向提供了新的视角和思路。

用于多元时间序列预测的图神经网络模型

现有用于多元时序预测的图神经网络模型大多基于预定义图以静态的方式捕捉时序特征,缺少对于系统动态适应和对时序样本之间潜在动态关系的捕捉.提出用于多元时序预测的图神经网络模型(MTSGNN).该模型在一个图学习模块中,采用数据驱动的方式学习时间序列数据的静态图和动态演化图,以捕捉时序样本之间的复杂关系.通过图交互模块实现静态图和动态图之间的信息交互,并使用卷积运算提取交互信息中的依赖关系.利用多层感知机对多元时序进行预测.实验结果表明,所提模型在6个真实的多元时间序列数据集上的预测效果显著优于当前最先进的方法,并且具有参数量较小、运算速度较快的优点.

多源知识融合的方面级情感分析模型

方面级情感分析(ABSA)是一项细粒度情感分析任务,其目的是针对评论语句中出现的特定方面给出对应的情感极性。现有的基于深度学习的ABSA方法大多侧重于评论语句语义和句法的挖掘,往往忽略了评论语句可能涉及的概念知识和情感程度信息。针对此问题,提出一种融合多源知识的神经网络模型,通过句法依赖揭示句子的结构框架、词共现捕捉单词之间的语义联系、情感网络和概念图谱的嵌入为模型提供情感和背景知识,共同实现评论语句上下文与评价方面的增强表示,并通过双交互注意力模式实现评论语句上下文与评价方面的协调优化。通过在4个公开数据集上的实验验证,该模型在ABSA任务中,准确率分别达到了75.00%、77.90%、81.55%、90.10%,与基准模型相比均有所提高。研究成果不仅验证了多源知识融合在ABSA任务中的有效性,也为未来的研究提供了新的思路和方法。

面向目标交互图神经网络的多模态方面级情感分析

对于多模态方面级情感分析任务,除了需要提取出文本和图像的表示,还需要将它们与方面语义信息相结合处理。然而,以往的相关方法对方面与文本和图像信息之间的交互处理不够充分,即使使用注意力机制建立起模态全局之间的关联,也难以在细粒度表达出它们的交互。为了充分进行多模态之间细粒度上的信息交互,提出一种面向目标交互图神经网络,围绕文本、图像和方面三者的关系建模,采用交叉注意力获取面向方面目标的文本和图像全局表示;建立多模态交互图,以连接不同模态的局部及全局表示节点;使用图注意力网络在粗细两个粒度上充分融合特征。在两个基准数据集上进行实验,结果表明该模型相比于仅使用注意力机制的模型,具有更佳的情感分类效果。

基于交互式多模型因子图的自适应组合导航算法

针对复杂城市环境下因外部干扰或传感器故障而引起的传统车载导航系统定位精度下降的问题,提出了一种基于因子图的交互式多模型车载导航算法。基于因子图优化算法建立了IMU/GNSS/LIDAR组合导航系统模型,引入了交互式多模型对子系统传感器量测进行建模并构建变量节点,利用模型概率更新来优化传感器权重,并依据因子图非线性优化和增量平滑理论实现车载导航系统的解算与更新。实验结果表明:相比于自适应因子图算法,所提算法在复杂城市环境下的定位精度提高了26.2%。

结合用户共同意图及社交关系的群组推荐方法

已有的群组推荐模型,在求解用户表示时大多比较单调且仅简单利用用户间的社交关系,使得用户表示不够准确,并且大都未考虑用户共同意图以及社交关系对群组偏好的影响,导致推荐的项目很难符合用户的需求。基于此,提出一种结合用户共同意图及社交关系的群组推荐模型(GR-UCISI)。首先构造用户-项目交互历史与社交关系相结合的用户意图分离模型,利用图神经网络采集每个用户的用户-项目交互以及社交关系信息,求解用户意图和项目表示;其次利用网络游走算法与K-means聚类算法将用户分组,结合用户群组、用户意图以及群组意图聚合过程获取群组共同意图表示;最后根据群组共同意图表示与项目表示得出群组推荐项目列表。该方法充分考虑到用户的个性以及群组成员间的共性对群组偏好的影响,同时结合社交关系缓解数据稀疏性问题,提升模型性能。实验结果表明,与9个对比模型中推荐效果最好的模型相比,在Gowalla数据集上,GR-UCISI的Precision和NDCG指标值分别提高3.01%和5.26%;在Yelp-2018数据集上,GR-UCISI的Precision和NDCG指标值分别提高2.96%和1.12%。

跨模态语义时空动态交互情感分析研究

针对传统情感分析中存在的模态间交互性差、时空特征融合度低的问题,建立了一种跨模态的语义时空动态交互网络。通过引入双向长短期记忆网络挖掘各模态的时间序列特征,加入自注意力机制强化模态内特征的权重赋值,将自动筛选出的特征矩阵送入图卷积神经网络进行语义交互。然后以时间戳为基础进行特征聚合,计算聚合层的相关系数,获得融合后的联合特征,实现跨模态空间交互,最终完成情感极性的分类与预测。使用公开数据集对所提出的模型进行评估验证,实验结果表明,多模态时间序列提取和跨模态语义空间交互机制可以实现模态内和模态间特征的全动态融合,有效地提升了情感分类的准确率和F1值,在CMU-MOSEI数据集上分别提高了1.7%~13.5%和2.1%~14.0%,表现出良好的健壮性和先进性。

考虑动态交互作用的智能车辆轨迹预测

在多车交互的动态场景中,智能车辆需要具备对周围车辆未来轨迹的预测能力,以实现安全高效行驶。本文提出一种考虑邻车动态交互作用的轨迹预测方法。首先基于目标车辆及周围车辆的历史轨迹信息,构建动态时空关联图,作为交互特征提取模块的输入,再运用图注意力机制获取历史时域上可变的交互特征参数;其次,将目标车辆历史时域信息与可变的交互特征参数融合,嵌入时间注意力机制得到上下文向量,再通过长短时记忆神经网络解码输出目标车辆的未来轨迹;最后,运用CitySim数据集对本文模型进行训练及验证,又采用CQSkyEye数据集对模型进行迁移性实验。结果显示:模型在5 s的预测时域上均方根误差为0.82 m,与其他预测模型的最优结果(0.96 m)相比,精度提升15%,并且可以提前2 s对车辆轨迹进行准确预测;对于迁移性能,本文模型相比其他模型有一定优势,在改变图构建的距离阈值参数后,5 s预测时域上的均方根误差为6.43 m,对比其他模型最优结果(12.41 m),精度提升48%。