http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Robust coefficients of correlation or spatial autocorrelation based on implicit weighting
Kalina Jan 한국통계학회 2022 Journal of the Korean Statistical Society Vol.51 No.4
Pearson product-moment correlation coefficient represents a fundamental tool for measuring linear association between two data vectors. In various applications, it is often reasonable to consider its weighted version known as the weighted correlation coefficient. This paper starts with theoretical considerations related to properties of the weighted correlation coefficient, particularly to its local robustness and relationship to other similarity measures. Inspired by the least weighted squares regression estimator, a robust correlation coefficient is investigated here together with its spatial autocorrelation extension. Finally, the considered methods are investigated in two image processing tasks.
Kalina YANEVA-NEDEVA 고려대학교 응용문화연구소 2018 에피스테메 Vol.0 No.20
The case of sudden stroke stroke aphasia presented in this chapter concerns Elisabeth, a French teacher (Yaneva-Nedeva, 2017). This study is part of the paradigm of general linguistics, language didactics, and theoretical and remedial neuro-psycholinguistics. Elizabeth is aphasic and she has motor sequelae, with some disorders of oral expression (sub-agrammatism) but a good verbal and written comprehension. The didactics of languages is approached, to show the interest of its application to the rehabilitation of the language of aphasic subjects, via the notion of "preceptorship" (Jacquet-Andrieu, 2012). Aphasia is a drama experienced by the person and his entourage. To offer aphasic patients care in respect of their human dignity, respecting medical ethics and without their being infantilized, we will draw on the variety of methods of language didactics. A relationship of trust is established in the care relationship over time that will explain how the linguist becomes a therapist and how he models the process of care through various methodologies, the purpose being to observe the strengths, linguistic progress at the end of the experiment. Le cas d’aphasie d’installation soudaine par accident vasculaire cérébral (AVC) présenté dans cet article concerne Élisabeth, professeur de français (Yaneva-Nedeva, 2017). Cette étude s’inscrit conjointement dans le paradigme de la linguistique générale, de la didactique des langues et de la neuro-psycholinguistique théorique et remédiative. Élisabeth est aphasique et elle garde des séquelles motrices, avec quelques désordres de l’expression orale (sub-agrammatisme) mais une bonne compréhension verbale et écrite. La didactique des langues est abordée, pour montrer l’intérêt de son application à la réadaptation du langage de sujets aphasiques, via la notion de « préceptorat » (Jacquet-Andrieu, 2012). L’aphasie est un drame humain, grave que subissent la personne et son entourage. Pour proposer aux personnes qui en sont atteintes une prise en charge dans le respect de leur dignité, en respectant l’éthique médicale et sans qu’elles se sentent infantilisées, nous puiserons dans la variété des méthodes d’enseignement de la didactique des langues. En éthique, dans toute relation de soin, un climat de confiance doit s’instaurer, au fil du temps de la remédiation cognitive ; nous expliquerons également comment le linguiste devient thérapeute et comment il modélise ce processus de soin, à travers divers curricula (ou programmes), le but étant d’observer les points forts, les progrès linguistiques à la fin de l'expérience.
칼리나(Kalina) 사단법인 한국언어학회 2021 언어학 Vol.- No.90
This paper aims to compare the participles of Korean and Solon Evenki, to identify similarities and differences in their semantic functions and usage. Solon Evenki has two participle forms as ‘-r’ and ‘-saa/-caa’, which are used to make an adnominal clause and a nominal clause. They both correspond to adnominal endings ‘-neun, -(eu)r, -(eu)n, -deon’ and nominalizing endings ‘-(eu)m, -gi’ of Korean. But in Korean, adnominal endings and nominalizing endings can be combined with verbs as well as adjectives. On the other hand, the participles of Solon Evenki can only combined with verbs. The participle forms of Solon Evenki, which correspond to ‘-neun’ and ‘-(eu)r’ of Korean, is ‘-r’. Unlike Korean, ‘-r’ does not mean an action is currently in progress. When indicating that an action is currently in progress, the progressive form ‘-ji-’ shall be combined before ‘-r’. The participle form of Solon Evenki, which correspond to the semantic function of ‘-(eu)n’ is ‘-saa/-caa’. The representative meaning of ‘-deon’ in Korean is interruption. When ‘-deon’ is used as interruption, Solon Evenki generally uses ‘-ji-saa’ which combines progressive form ‘-ji-’ and ‘-saa’, or uses ‘-saa/-caa’. The nominal ending ‘-(eu)m’ in Korean corresponds to ‘-r’ of Solon Evenki. But ‘-at/eosseum’(‘-았/었음’) which combined with ‘-at/eot-’(‘-았/었-’) corresponds to ‘-saa/-caa’ of Solon Evenki. The participle form of Solon Evenki, which corresponds to ‘-gi’, is also ‘-r’. ‘-(eu)m’ and ‘-gi’ in Korean can end sentences, but ‘-r’ in Solon Evenki has no such usage, it is expressed by the indicative ending.
Improving Adversarial Domain Adaptation with Mixup Regularization
Bayarchimeg Kalina,Youngbok Cho The Korea Institute of Information and Commucation 2023 Journal of information and communication convergen Vol.21 No.2
Engineers prefer deep neural networks (DNNs) for solving computer vision problems. However, DNNs pose two major problems. First, neural networks require large amounts of well-labeled data for training. Second, the covariate shift problem is common in computer vision problems. Domain adaptation has been proposed to mitigate this problem. Recent work on adversarial-learning-based unsupervised domain adaptation (UDA) has explained transferability and enabled the model to learn robust features. Despite this advantage, current methods do not guarantee the distinguishability of the latent space unless they consider class-aware information of the target domain. Furthermore, source and target examples alone cannot efficiently extract domain-invariant features from the encoded spaces. To alleviate the problems of existing UDA methods, we propose the mixup regularization in adversarial discriminative domain adaptation (ADDA) method. We validated the effectiveness and generality of the proposed method by performing experiments under three adaptation scenarios: MNIST to USPS, SVHN to MNIST, and MNIST to MNIST-M.