学科服务 更多>>

关键词： Facial expression   Valence   Arousal   HRC   AffectNet   AFEW   Aff-Wild  

摘要： We address the problem of facial expression analysis. The proposed approach predicts both basic emotion and valence/arousal values as a continuous measure for the emotional state. Experimental results including cross-database evaluation on the AffectNet, Aff-Wild, and AFEW dataset shows that our approach predicts emotion categories and valence/arousal values with high accuracies and that the simultaneous learning of discrete categories and continuous values improves the prediction of both. In addition, we use our approach to measure the emotional states of users in an Human-Robot-Collaboration scenario (HRC), show how these emotional states are affected by multiple difficulties that arise for the test subjects, and examine how different feedback mechanisms counteract negative emotions users experience while interacting with a robot system.

关键词： Duskin nerve   2-category   Suspension 2-category   Joyal's cell category  

摘要： We give an explicit and purely combinatorial description of the Duskin nerve of any (r + 1)-point suspension 2-category, and in particular of any 2-category belonging to Joyal's cell category circle minus(2). (C) 2020 Elsevier B.V. All rights reserved.

关键词： Memory   Judgments   Decision-making   Category  

摘要： People are very good at making relative judgments (determining where a stimulus ranks in a distribution) but not very good at making absolute judgments (determining the absolute properties of the stimulus independent of context). Thus, when evaluating various stimuli, people tend to encode ordinal rank rather than objective quality. This leads to a robust bias in memory-based judgments: when the nature of the distribution shifts, people fail to update their memories of ordinal rank, and thus incorrectly judge a stimulus that is at the top [bottom] of the distribution at Time 1 to also be at the top [bottom] of a different distribution at Time 2. In two studies, we demonstrate a crucial moderator of the relative encoding bias on memory-based judgments: category saliency at the time of encoding. More specifically, we reveal that this memory-based judgment bias is reduced if it is made salient at the time of encoding that the stimuli from Time 1 and Time 2 are from different categories and thus different distributions.

关键词： Store Layouts   Retail Shelf Effects   Multi-Category Demand Models   Purchase Incidence Decisions  

摘要： Grocery stores typically contain hundreds of product categories. Physically arranging these product categories among a limited number of store aisles and shelves is a challenging task. We propose a multi-category demand model that considers physical locations of product categories to be important determinants of consumers' purchase incidence decisions in these categories at the store, and examine two issues: (1) to what extent does store layout influence shopping basket decisions? (2) how should the store layout be re-designed to achieve optimal profit? Our analyses suggest the retailer can significantly increase store profitability by re-organizing retail layout configurations.

摘要： Given a differentially graded (DG)-category A, we introduce the bar category of modules (Mod) over bar -A. It is a DG enhancement of the derived category D(A) of A, which is isomorphic to the category of DG A-modules with A(infinity)-morphisms between them. However, it is defined intrinsically in the language of DG categories and requires no complex machinery or sign conventions of A(infinity)-categories. We define for these bar categories Tensor and Hom bifunctors, dualisation functors, and a convolution of twisted complexes. The intended application is to working with DG-bimodules as enhancements of exact functors between triangulated categories. As a demonstration, we develop a homotopy adjunction theory for tensor functors between derived categories of DG categories. It allows us to show in an enhanced setting that given a functor F with left and right adjoints L and R, the functorial complex FR (FactR) under right arrow FRFR (FRtr-trFR) under right arrow FR (tr) under right arrow Id lifts to a canonical twisted complex whose convolution is the square of the spherical twist of F. We then write down four induced functorial Postnikov systems computing this convolution.

关键词： Vertebrae recognition   Object detection   Message passing   Dictionary learning   Label consistency  

摘要： Accurate vertebrae recognition is crucial in spinal disease localization and successive treatment planning. Although vertebrae detection has been studied for years, reliably recognizing vertebrae from arbitrary spine MRI images remains a challenge. The similar appearance of different vertebrae and the pathological deformations of the same vertebrae makes it difficult for classification in images with different fields of view (FOV). In this paper, we propose a Category-consistent Self-calibration Recognition System (Can-See) to accurately classify the labels and precisely predict the bounding boxes of all vertebrae with improved discriminative capabilities for vertebrae categories and self-awareness of false positive detections. Can-See is designed as a two-step detection framework: (1) A hierarchical proposal network (HPN) to perceive the existence of the vertebrae. HPN leverages the correspondence between hierarchical features and multi-scale anchors to detect objects. This correspondence tackles the image scale/resolution challenge. (2) A Category-consistent Self-calibration Recognition (CSRN) Network to classify each vertebra and refine their bounding boxes. CSRN leverages the dictionary learning principle to preserve the most representative features;it imposes a novel category-consistent constraint to force vertebrae with the same label to have similar features. CSRN then innovatively formulates message passing into the deep learning framework, which leverages the label compatibility principle to self-calibrate the wrong pre-recognitions. Can-See is trained and evaluated on a capacious and challenging dataset of 450 MRI scans. The results show that Can-See achieves high performance (testing accuracy reaches 0.955) and outperforms other state-of-the-art methods. (C) 2020 Elsevier B.V. All rights reserved.

关键词： Mineral water   Medicinal water   Bottled water brand authentication   Isotopic quality preservation  

摘要： The hydrogen (delta H-2) and oxygen (delta O-18) isotopic composition of 45 Polish brands of bottled waters was analysed and evaluated in the context of two areas of its potential application in bottled water industry: (i) the origin authentication, and (ii) the quality monitoring in context of preservation of natural physicochemical parameters and isotopic composition. Three main market categories of bottled waters were analysed: natural mineral waters, spring waters and medicinal waters. The results obtained showed that isotopic composition for domestic bottled waters varied in wide range from-11.0%o to-0.6%o for delta O-18 and from -76.8 parts per thousand to -50.2 parts per thousand for delta H-2. The bottled waters generally preserve the natural isotopic composition of their source of origin and discharge. However, the influence of the bottling process on the natural isotopic composition of water may be observed in relation to medicinal waters of higher salinity and natural CO2 content. The physicochemical parameters and isotopic composition of waters stored in PET-made bottles may change over time indicating deterioration of the water quality. The application of H and O isotopic composition for bottled water brand authentication among natural mineral waters and spring waters is very limited due to similar climatic factors controlling formation of meteoric groundwaters in hydrogeological systems in local or global scale. However, it is possible to differentiate and authenticate bottled waters which have specific isotopic composition resulting from unique origin connected first of all with geological factors like water-rock interaction or admixture of water of non-meteoric origin. Among Polish brands of bottled waters such unique H and O isotopic composition is associated usually with some medicinal waters.

关键词： cytology   fine needle biopsy   thyroid neoplasms  

摘要： Objective To compare the frequency of fine needle aspiration (FNA) cytological results between nodules 1 cm or less and nodules greater than 1 cm. Methods All FNAs performed between January 2016 and December 2019 at an institute in Brazil were analysed. For each nodule, at least two conventional slides were produced (one stained by Giemsa and the other by hematoxylin and eosin). All cases were reviewed by two cytopathologists and were reported following the Bethesda System. Clinical information (gender and age) and ultrasound data (nodule size and location) were collected. The magnitude of association was measured using the prevalence ratio (PR) with 95% confidence intervals (CIs) considering two groups of nodules: 1.0 cm or less, and greater than 1 cm. Results A total of 3703 nodules were analysed from 3265 patients (2906 women [88.48%], 359 men [11.51%], with a combined mean age of 52 years). Considering the prevalence ratio of Bethesda categories between these two groups of nodules, the nondiagnostic or unsatisfactory category (PR: 3.0, 95% CI: 2.2-4.2) and the suspicious for malignancy category (PR: 1.6, 95% CI: 1.1-2.4) were significantly associated with nodules measuring 1 cm or less. Conclusion Our results demonstrated that nodules 1 cm or less are significantly associated with the nondiagnostic and suspicious for malignancy categories of the Bethesda System when compared to nodules greater than 1 cm.

关键词： dcpo半群   Cartesian闭范畴   自由dcpo半群  

摘要： 本文定义了一类特殊的序结构,称为dcpo半群.该结构是一个带有关于Scott拓扑连续的半群运算的定向完备偏序集.传统的各类幂domain都是特殊的dcpo半群.本文讨论了dcpo半群的子代数、和、乘积等结构,证明以Scott连续函数为态射的dcpo半群范畴是Cartesian闭范畴.特别地,本文讨论了dcpo的自由dcpo半群,利用抽象基的方法构造出连续domain的自由dcpo半群,并证明该自由dcpo半群关于连续性、代数性、可数基等封闭.

关键词： 《诗经》   非范畴化   反复辞格   翻译  

摘要： 在文学翻译领域,诗歌翻译非常具有挑战性。《诗经》为中国第一部诗歌总集,其翻译研究备受学界关注。反复辞格是《诗经》语言表达的一个显著特征,在表意和修辞上具有显著效果。传统反复辞格翻译倾向于对反复辞格的类型、功能,及文学、美学理论下反复翻译策略的探究,较少涉及该辞格翻译的认知过程,其研究维度与内容的单一性一定程度上局限了《诗经》语言英译研究的广度与深度。非范畴化不仅是语言变化发展的重要途径,更是人类认识的一种重要方式。其重视语言、文化、思维等综合因素在翻译认知过程中的作用,实现语言实体在功能与意义上的变化。鉴于此,本研究尝试从认知理论之非范畴化视角出发,对《诗经》中反复辞格翻译进行探究。本文基于非范畴化理论,以《诗经》中反复辞格为研究对象,从非范畴化理论的四大特征出发,对反复辞格翻译的非范畴化过程进行细致解读。文章通过对《诗经》原文及理雅各、许渊冲的《诗经》英译本:The She King与Book of Poetry中的反复辞格进行分类、描写与阐释,归纳总结了《诗经》英译中反复词汇、句法结构的非范畴化特征,并尝试论述了在反复翻译过程中所体现的内在动因与认知机制。研究发现:(1)《诗经》中反复辞格的特点主要表现在词汇和句法结构两个维度。词汇反复可被划分为四类:象声词、形容词、动词和名词,且每类在《诗经》中的使用频次和所占比存在较大差异。其中,形容词性叠词在原文本中的使用频率最高。句法结构的反复可被划分为两类:回环结构和递进结构,通过对原诗中反复句法的分类统计发现,相对递进结构而言,回环结构的使用频率较高。(2)本文通过对《诗经》英译本中反复词汇非范畴化特征的分析,将其总结为五大类:第一,形容词-动词类,语义上,由描述义向动态义转变,实现语义的动态化。句法形态上,此类词可在小句中充当谓语,具有时态、体态等变化,或与某种介词搭配构成固定短语。第二,名词-动词类,语义上,失去其原有的指称意义,语义实现抽象和泛化。句法形态上,丧失原本的句法限制,不再具有名词词类的形态分布特征。第三,名词-形容词类,语义上,失去原有的指称义,用以描述事物的样态特征。句法形态上,不再具有名词的句法分布特征。第四,动词-名词类,语义上,由动态义转变为指称义。句法形态上,丧失其范畴属性特征,不再具有动词原型的人称、时、体的变化。第五,动词-形容词类,语义上,由动态义向描述义转变。句法形态上,不再具有动词词类的形态分布特征,丧失原本的句法限制。(3)本文通过句法结构层面的反复翻译分析发现,其往往通过打破常规句法的相对稳定性,破坏词汇搭配的规则来实现句法结构的非范畴化。反复句法虽丧失了常规语法特征,但其获得了新的功能和意义。(4)语言的经济性原则和范畴的动态属性特征是《诗经》中反复翻译非范畴化的内在动因,隐喻和识解是实现这一非范畴化过程的认知机制。概念隐喻和范畴隐喻较充分地解释了反复词类的转换过程,Object-to-Quality和Quality-to-Activity是其主要涉及的范畴隐喻。本文在认知视域内对《诗经》中反复辞格翻译进行了探究,总结了反复辞格的非范畴化特征及认知机制,希望本研究能丰富反复辞格翻译的研究视角,并为其它诗歌修辞的翻译研究及实践提供微小的借鉴意义。