In sensor-based human activity recognition (HAR), data from different people are usually assumed to be identically independently distributed. However, an activity may vary from person to person so that a model trained on one person may perform terribly on other people because of different distributions, AKA domain shift. To solve the problem, we propose a domain-specific mixture of experts (DSME), especially for generalizing sensor-based HAR. First, DSME enables several experts to learn from their responsible source domains, respectively. Second, a domain-sniff module is used to extract features from each source domain to learn weights for experts, which manages to make classification more accurate. Finally, extensive experiments on four public datasets show that DSME gains 0.82%, 4.07%, 1.56%, and 0.25% F1-score improvement compared with the-state-of-art baselines in the daily and sports activities (DSA), physical monitoring activity for aging people 2 (PAMAP2), OPP, and the University of California, Irvine, HAR (UCIHAR) datasets. Besides, visualization analysis is also conducted to further prove that DSME can identify the discrepancy among training domains and then relieve domain shift in the best way.

Sensor-based human activity recognition is important in daily scenarios such as smart healthcare and homes due to its non-intrusive privacy and low cost advantages, but the problem of out-of-domain generalization caused by differences in focusing individuals and operating environments can lead to significant accuracy degradation on cross-person behavior recognition due to the inconsistent distributions of training and test data. To address the above problems, this paper proposes a new method, Multi-channel Time Series Decomposition Network (MTSDNet). Firstly, MTSDNet decomposes the original signal into a combination of multiple polynomials and trigonometric functions by the trainable parameterized time series decomposition to learn the low-rank representation of the original signal for improving the extraterritorial generalization ability of the model. Then, the different components obtained by the decomposition are classified layer by layer and the layer attention is used to aggregate components to obtain the final classification result. Extensive evaluation on DSADS, OPPORTUNITY, PAMAP2, UCIHAR and UniMib public datasets shows the advantages in classification accuracy and stability of our method compared with other competing strategies, including the state-of-the-art ones. And the visualization is conducted to reveal MTSDNet’s interpretability and layer-by-layer characteristics. Note to Practitioners —This paper is motivated by solving the problem of decreased classification accuracy caused by differences in human activity, such as the differences in walking patterns between young and elderly people. Models trained on young people’s data are difficult to accurately recognize the activities of the elderly. This is a problem of domain generalization, where the distribution differences between training and testing data require the model’s ability to generalize across domains. We propose a model based on time series decomposition, which helps the model learn universal features and effectively improve its generalization ability. This method can be further extended to time series tasks with varying domain distributions.

根据对现有数份书目的统计分析,韩国所藏中国刊本通俗小说总量可观,但其主体为清中后期坊刻本及清末民国时期铅石印本,学术文献价值相对较高的明版及清早期刊本藏量有限.不过,其中仍不乏孤善之本,既有《型世言》《莽男儿》这样的中国亡佚小说,也有《皇明英烈传》《百家公案》《廉明公案》《醒世姻缘传》《引凤箫》等明清小说的早期重要版本,足资学术.而在中国通俗小说传入朝鲜半岛的过程中,朝鲜王室及燕行使发挥了举足轻重的历史作用.虽说在儒学占据社会文化主导地位的朝鲜半岛中国通俗小说整体上仍难以登入大雅之堂,但凭借汉字文化的特殊力量,它们终究在或明或晦的时空里,影响了一代又一代的朝鲜人.

针对当前师范类高校在人工智能教育人才培养过程中如何契合新工科交叉融合实践能力培养内涵的问题,对于产教融合的思维素养标准重定义、实践课程体系重构、实践平台建立、实践质量评估4个方面的建设提出路径,以上海师范大学教育部新工科实践项目的人工智能相关专业为例,介绍师范类高校人工智能教育人才实践能力培养体系构建过程,最后说明取得的实践探索成效.

The formation dip angle is an important characteristic parameter that reflects underground structures, and it is widely used in seismic exploration and geological interpretation. However, the routine dip angle calculation method suffers from the problems of artifact interference and insufficient accuracy, which restricts the development of fine exploration technology. More accurately obtaining the formation dip angle has become a general concern. To solve these problems, this paper proposes a method for calculating the formation dip angle using deep learning. The dip angle calculation is considered a regression problem. By establishing a database of synthetic seismic data and dip data tags, data-driven fitting of nonlinear functional relationships between the seismic data and dip angles are achieved, and intelligent dip calculations are realized. The method is verified using the synthetic sedimentary model and actual data and is compared with the mainstream dip angle calculation method in the industry. The results show that this method more realistically reflects the undulating characteristics of a structure with both high accuracy and anti-interference ability.

Chinese Named Entity Recognition (NER) has received extensive research attention in recent years. However, Chinese texts lack delimiters to divide the boundaries of words, and some existing approaches cannot capture the long-distance interdependent features. In this paper, we propose a novel end-to-end model for Chinese NER. A new global word boundary detection approach is designed to capture the semantic dependency via a self-attention mechanism to represent character embedding by assigning compatible weights for each character in a sentence. To improve the representation ability of Chinese named-entity boundaries, we introduce position-aware influence propagation with the Gaussian kernel for each character, which combines convergence propagation and radiation propagation. Convergence propagation mainly measures the influence of surrounding characters on the target character. The purpose of radiation propagation is to measure the range of influence of the target character on surrounding characters. The proposed method has been evaluated and shown to offer strong performance in two Chinese NER datasets: MSRA and PFR.

古典小说文本往往配插有一定数量的图像,或为情节性插图,或为人物绣像,这成为中国古代小说的一个民族特点.而以《三国志演义》《水浒传》《西游记》《红楼梦》为代表的四大名著,无论从图像的数量质量还是样式风貌,均发挥着特殊的典范和引领作用.其中,《三国志演义》《水浒传》《西游记》存世明刻本颇夥,图像多为情节性插图,主要包括福建建阳刻本的上图下文与江南书坊刻本的整版大幅两大类型,入清之后,其图像则以人物绣像为主,至清季石印技术传入,情节性插图才重新回到文本;与上述三种小说不同,《红楼梦》在木刻本时代,没有出现过情节性插图,均为人物绣像,这一遗憾,直到晚清民国时期始得弥补,故《红楼梦》近代石印图像本的绘印甚为繁荣,是四大名著中最为丰富的.小说图像蕴含着宽广的学术研究空间,小说图像学正得以蓬勃展开,方兴未艾.

昔时广州东门外(今先烈东路北侧)有一处现已消失的景观——金娇墓,墓主金娇为清末广州名妓,于宣统元年(1909)正月初九日大沙头火灾中不幸罹难,好事者将其舁葬沙河,树碑立传,士女游观,不绝于道.宣统二年(1910)五月,广州惜花社出版了梁纪佩《金娇墓》小说,这是金娇故事的首次文学书写,凡十三回,叙写了金娇的身世故事以及金娇墓的营造始末,小说亦体现了作者对于史实与虚构的特殊处理方式.自清末以降,金娇墓的各种文学书写渐次产生,横跨诗词、文、粤讴、小说等文类,又因附近黄花岗烈士墓的建成,呈现出"烈士美人"的特殊意象组合.20世纪三四十年代,随着金娇墓实物景观的凋零,其文学书写中的岭南地方史事痕迹开始淡化,转向接入更为普泛的古典小说戏曲叙事传统.

The PM2.5 index is a vitally important air pollution indicator that reflects the atmospheric concentration of particulate matter with diameters less than 2.5 mu m. Recently, researchers have sought to address the inadequacies involved with the capture of environmental data from monitoring stations by developing deep learning methods for estimating PM2.5 concentrations in real time based on image data. However, the currently available methods neglect the effect of PM2.5 concentrations in earlier periods on the current air quality. The present work addresses this issue by proposing a novel hybrid model that first applies the DenseNet model to extract the local high-order spatial features of images captured over several hours or days at a fixed location, and then feeds the extracted features as a time series into an attention-based long short-term memory model. Ultimately, a fully connected layer is applied to establish a regression model to evaluate the PM2.5 concentrations reflected in the image data. The effectiveness of the proposed method for assessing PM2.5 concentrations is demonstrated by its application to two datasets composed of images captured in Shanghai and Beijing, China.

Accurate air pollutant concentrations prediction allows effective environment management to reduce the impact of pollution. The encoder-decoder model based on long short-term memory (LSTM) demonstrated great potential in air pollutant concentrations prediction. However, the influence of the hidden vector on the output of the decoder at each moment may be different during the long time-series prediction problem. In this paper, an attention mechanism is introduced in the decoder part to further improve the final prediction of the pollutant concentrations by filtering out some noise. In the experimental stage, we exploited the data collected from five representative cities in North China (Beijing, Tianjin, Shijiazhuang, Taiyuan, and Baotou) from 2014 to 2019 as the experimental dataset, and added the auxiliary data unique to North China. Then we divided the datasets into four seasonal datasets (spring, summer, autumn, and winter) to obtain targeted seasonal prediction models for the different seasons. The experimental results show that the model can predict the future trend of the pollutant concentration in a certain season relatively accurately.

俗文学文献由印刷本及抄写本构成,它们总体上缺乏经典性,又受艺术体制特点的影响,其文本呈现出较为显著的流动性,并对古典文献学产生若干学术上的不适应,甚至学理上的冲突.故有必要在实践基础上,探索构建既取法古典文献学、又具有一定独立性和指向性的"俗文学文献学".至于如何构建"俗文学文献学"?笔者认为可以参照古典文献学的"三位一体"构成,结合俗文学文献的学术特性和实际情况,分别从目录学、版本学、校勘学、流通学四大块面,展开相应的理论探索与学术实践.

Air pollution has become one of the critical environmental problem in the 21st century and has attracted worldwide attentions. To mitigate it, many researchers have investigated the issue and attempted to accurately predict air pollutant concentrations using various methods. Currently, deep learning methods are the most prevailing ones. In this paper, we extend a comprehensive review on deep learning methods specifically for air pollutant concentration prediction. We start from the analysis on non-deep learning methods applied in air pollutant concentration prediction in terms of expertise, applications and deficiencies. Then, we investigate current deep learning methods for air pollutant concentration prediction from the perspectives of temporal, spatial and spatio-temporal correlations these methods could model. Further, we list some public datasets and auxiliary features used in air pollutant prediction, and compare representative experiments on these datasets. From the comparison, we draw some conclusions. Finally, we identify current limitations and future research directions of deep learning methods for air pollutant concentration prediction. The review may inspire researchers and to a certain extent promote the development of deep learning in air pollutant concentration prediction.

Air pollution has become a critical issue in human’s life. Predicting the changing trends of air pollutants would be of great help for public health and natural environments. Current methods focus on the prediction accuracy and retain the forecasting time span within 12 hours. Shorter time span decreases the practicability of these perditions, even with higher accuracy. This study proposes an attention and autoencoder (A&A) hybrid learning approach to obtain a longer period of air pollution changing trends while holding the same high accuracy. Since pollutant concentration forecast highly relates to time changing, quite different from normal prediction problems like autotranslation, we integrate “time decay factor” into the traditional attention mechanism. The time decay factor can alleviate the impact of the value observed from a longer time before while increasing the impact of the value from a closer time point. We also utilize the hidden states in the decoder to build connection between history values and current ones. Thus, the proposed model can extract the changing trend of a longer history time span while coping with abrupt changes within a shorter time span. A set of experiments demonstrate that the A&A learning approach can obtain the changing trend of air pollutants, like PM2.5, during a longer time span of 12, 24, or even 48 hours. The approach is also tested under different pollutant concentrations and different periods and the results validate its robustness and generality.

孙逊教授的学术生涯,肇始于20世纪70、80年代的《红楼梦》研究,红学是他一生的学术底色.之后,他又先后驰骋于古典小说艺术理论、小说与宗教、小说与城市、东亚汉文小说研究诸领域,在文本阐释、理论探索、文化考察和文献整理等方面,均取得了丰硕的学术成果.他是新时期作出重要学术贡献的代表性红学家之一,是一位富有开拓精神的古代小说研究名家,也是国内东亚汉文小说研究的倡导者和推进者,其学术人生多有启迪后学之处.

作为第一部影印出版的程本《红楼梦》,台北青石山庄本发挥过颇为重要的学术作用,但其收藏者及出版者"胡天猎"却始终不为人所知.根据新发现的信札,"胡天猎"真名韩镜塘,久居辽北,曾任东北大学工科教授,1948年赴台湾,任教台北多家高校.韩氏喜爱古小说,从国内旧书肆及日本东京文求堂,搜访购藏小说善本,并以一己之力编印青石山庄"古本小说丛书",可惜仅影印了《红楼梦》《三国志演义》等三种便中途而辍.其小说藏书精华,也于1967年出售给美国耶鲁大学东方图书馆.韩镜塘的名字不应湮没,他有资格在古代小说学术史上写下光耀的一笔.

《崔炜》是晚唐文人裴铏著名小说集《传奇》中的一篇,历来所受学界关注不够,但它实际上是一篇具有特殊价值的作品.《崔炜》乃裴铏"有意为小说"的产物,从故事背景的"有意"叠加,到情节模式的"有意"缀合,再到结构叙事的"有意"处置,《崔炜》取得了艺术成功,但也存在明显不足.裴铏的这一创作实例,有助于我们更加深细地理解唐人"有意为小说"命题的内涵和外延,若就此而言,《崔炜》具有学术上的经典个案意义.

明清时期中国和朝鲜半岛保持了较为稳定的外交关系,抵达中国的朝鲜使团数量甚是可观.因口语交流不畅,朝鲜文人多假借笔谈方式与中国文人展开交流,留下了丰富的笔谈资料.近年新发现的中国文人刘大观在1799年与朝鲜使者徐滢修、韩致应之问的笔谈稿手卷,不同于收录于“朝天录”“燕行录”文献中的笔谈整理稿,而是珍贵的笔谈原纸.该手卷不仅有助于揭示中朝文人在进行笔谈时的用纸、书写方式、敏感话题的涂抹毁弃、时间题署等笔谈稿形制,还可以借此考察笔谈稿的流传、誊录编集情况.需要指出的是,作为一种记录口头交谈的特殊史料,笔谈文献虽有重现交流现场的作用,但它大多经过事后的增删整理,研究利用时应予注意并酌情处理.

1981年4月24日至5月14日,日本红学家伊藤漱平与他的老师松枝茂夫应邀访问中国,与诸多中国红学家会晤交流.1982年1月,伊藤漱平致信曾保泉,言及访问细节以及若干红学史的故人旧事;随信又附赠了他私人特制贺年卡,印有红学随笔一篇,中有题咏林黛玉的日本俳句及中国词作《甘州曲·忆潇湘妃子》,伊藤漱平以“人未瘦”“为花忙”为核心话语,文学地阐述了其对黛玉、宝玉精神品格的独特理解.伊藤漱平曾自号“泥人”“伊泥卿”“红梦楼主”,书斋名为“两红轩”,一生醉心于《红楼梦》的翻译与研究,值其逝世十周年之际,谨以此文表达对他的缅怀和纪念.

明余象斗编撰《皇明诸司廉明奇判公案》,是明代诸司体公案小说集的第一部,存世版本达九部.最新发现的朝鲜燕行使旧藏明末金陵大业堂刊本,独家保存着一篇曾因触犯地方权贵而被抽毁的小说《王巡道察出匿名》,它揭开了一段隐藏在书叶背后的明代小说出版史秘闻.以此为版本标记物,考察其在诸版本中的存删痕迹,可以厘清存世九部版本的学术关系.同时《廉明公案》在东亚的流播史,从书籍史和小说史的角度,带给研究者新的思考,海外存藏汉籍对于明代小说研究的特殊文献意义,仍需深入认知开掘;明代公案小说的时事因子和现实品格,有待进一步关注探讨;而公案小说的文体性质与小说史价值,亦可在小说知识学的维度下,获得新的观照、评估和阐释.