Expression Recognition Method Using Improved VGG16 Network Model in Robot Interaction

Aiming at the problems of poor representation ability and less feature data when traditional expression recognition methods are applied to intelligent applications, an expression recognition method based on improved VGG16 network is proposed. Firstly, the VGG16 network is improved by using large convolution kernel instead of small convolution kernel and reducing some fully connected layers to reduce the complexity and parameters of the model. Then, the high-dimensional abstract feature data output by the improved VGG16 is input into the convolution neural network (CNN) for training, so as to output the expression types with high accuracy. Finally, the expression recognition method combined with the improved VGG16 and CNN model is applied to the human-computer interaction of the NAO robot. The robot makes different interactive actions according to different expressions. The experimental results based on CK + dataset show that the improved VGG16 network has strong supervised learning ability. It can extract features well for different expression types, and its overall recognition accuracy is close to 90%. Through multiple tests, the interactive results show that the robot can stably recognize emotions and make corresponding action interactions.

Arabic Learners’ Production of Regular English Past Tense Verbs

Previous studies show that second language (L2) learners of English sometimes produce the verb with proper past tense inflectional morphology as in help[t] and sometimes repair the cluster, as in helpø or hel[pəd]. Complicating matters, these studies focused on L2 learners whose native languages disallowed codas or had very restricted codas. Thus, it is difficult to tell whether any problems in producing past tense morphology are due to first language L1-transferred coda restrictions, or an inability to acquire the abstract feature of past tense. To rule out native language syllable structure interference, this paper aims to examine the production of the English regular past tense verb by Arabic L1 ESL learners, a language that allows complex codas. The paper also examines the role of a phonological universal, the Obligatory Contour Principle (OCP) that disallows two adjacent similar sounds, and its effect on learners’ production. The data come from twenty-two English as a Second Language (ESL) students at three levels of proficiency. The task was a sentence list eliciting target clusters in past tense contexts that violate manner in OCP: fricative + stop ([st], [ft]) vs. stop + stop ([pt], [kt]). Results show that L1 Arabic speakers have difficulty in producing past tense morphology, even though their L1 allows complex codas. Fricative + stop clusters are repaired (epenthesis/deletion) at a lower rate (low =25.71%, intermediate = 6.6%, high=11.11%) than stop + stop clusters (low=57.14%, intermediate = 40.27%, high=22.91%). The higher rate of repair is clear in stops + stop clusters suggesting that learners abide by phonological universals and prefer not to violate OCP. Finally, proficiency level has an effect on target-like production, as higher-proficiency learners produce past-tense morphology at a higher rate than lower-proficiency learners. Together, these results indicate that L1 transfer is not the only source of difficulty in the production of past tense morphology, and that the abstract feature of tense is problematic, particularly at the early stages of ESL development.

Quoi-sluices in French

Sluicing has traditionally been analyzed as an operation involving wh-movement and deletion (Merchant 2001). French is a language that has both fronted and wh-in situ strategies; on the surface, however, it seems that French sluices do not involve (overt) movement, in spite of this being an available option. For nearly all wh-words, the in situ and moved forms are the same; the exception is que/quoi ‘what’— que is found in fronted wh-questions alone, while quoi is found in situ. In sluicing, only quoi surfaces, suggesting that French may be a challenge for the movement-and-deletion approach (Dagnac 2019). By formalizing an analysis within a late insertion approach to the syntax-morphology interface, I argue that not only do sluices in French involve full structure, but that they involve movement as well. I assume that the wh-word is initially represented in the syntactic derivation as an abstract feature bundle. The morphological form is determined in the mapping of syntax to morphology by locality-dependent Vocabulary Insertion (VI) rules that are sensitive to C. These rules apply only after ellipsis occurs. Following Thoms (2010), I argue that C is targeted in sluicing, and as a result destroys the context that would trigger que. This analysis is able to capture sluicing in French, while explaining the behavior of quoi more generally.

Accent Recognition with Hybrid Phonetic Features

The performance of voice-controlled systems is usually influenced by accented speech. To make these systems more robust, frontend accent recognition (AR) technologies have received increased attention in recent years. As accent is a high-level abstract feature that has a profound relationship with language knowledge, AR is more challenging than other language-agnostic audio classification tasks. In this paper, we use an auxiliary automatic speech recognition (ASR) task to extract language-related phonetic features. Furthermore, we propose a hybrid structure that incorporates the embeddings of both a fixed acoustic model and a trainable acoustic model, making the language-related acoustic feature more robust. We conduct several experiments on the AESRC dataset. The results demonstrate that our approach can obtain an 8.02% relative improvement compared with the Transformer baseline, showing the merits of the proposed method.

Development of a Wilks feature importance method with improved variable rankings for supporting hydrological inference and modelling

Feature importance has been a popular approach for machine learning models to investigate the relative significance of model predictors. In this study, we developed a Wilks feature importance (WFI) method for hydrological inference. Compared with conventional feature importance methods such as permutation feature importance (PFI) and mean decrease impurity (MDI), the proposed WFI aims to provide more reliable variable rankings for hydrological inference. To achieve this, WFI measures the importance scores based on Wilks Λ (a test statistic that can be used to distinguish the differences between two or more groups of variables) throughout an inference tree. Compared with PFI and MDI methods, WFI does not rely on any performance measures to evaluate variable rankings, which can thus result in less biased criteria selection during the tree deduction process. The proposed WFI was tested by simulating monthly streamflows for 673 basins in the United States and applied to three interconnected irrigated watersheds located in the Yellow River basin, China, through concrete simulations for their daily streamflows. Our results indicated that the WFI could generate stable variable rankings in response to the reduction of irrelevant predictors. In addition, the WFI-selected predictors helped random forest (RF) achieve its optimum predictive accuracy, which indicates that the proposed WFI could identify more informative predictors than other feature importance measures.

The word formation potential of politicians’ surnames on the example of the category of names of abstract features

Research has shown that in contemporary political communication there are often derivatives belonging to the category of abstract feature names derived from the names of politicians. In total, I amassed 166 such units. I analyze their word-formation structure and the meaning that results not only from morphology, but is also contextually conditioned. The most productive is the suffix -izm/-yzm (gierkizm, schetynizm, balcerowizm) and the suffix -izna/-yzna (komorowszczyzna, michnikowszczyzna). Formations formed with the help of other suffixes are definitely less numerous, e.g. -ość (kaczyńskość), -ada (kukiziada) or complex formants: -aizm (dudaizm), -stwo (kaczorstwo), -ostwo (pawłowiczostwo), - arstwo (palikociarstwo) and -olioza (palikotolioza). Other derivative mechanisms are used to create odantroponimic names of ideology and a set of views, e.g. paradigmatic derivation with the disintegration of the topic (olszewia) or complex derivation (korwinomania, wałęsocentryzm). The derivatives in question are most often used by people who do not accept the phenomena signed by these words. The high frequency of names is undoubtedly influenced by new forms of communication, especially the Internet, and the mediatization of politics.

Kant, Eudaimonism, Act-Consequentialism and the Fact of Reason

Kant considers eudaimonism as his main opponent and he assumes that his ethics is the only viable alternative to eudaimonism. He does not explicitly address theories differing from both eudaimonism and from his own. I argue that whilst Kant and Act-Consequentialists advocate different normative principles, their positions share the important abstract feature that they establish what is to be done from a rational principle and not based on what is in the self-interest of the respective agent, as Kant thinks eudaimonism does. Act-Consequentialism is thus closer to Kant’s ethics than is often assumed. I will demonstrate and vindicate this point with a new interpretation of the Fact of Reason. This reading also establishes that the notion of a Fact of Reason is less contentious than many of Kant’s critics believe. We should not expect that the Fact establishes Kantianism. Instead, the Fact is only supposed to count against a specific competing view of morality, namely, eudaimonism. Act-Consequentialists can accept the Fact as well.

Semantico-Functional Features of Expressive Derivatives in the Artistic Discourse of V. Makanin

The paper is devoted to the study of semantic and functional features of expressive derivatives, both usual and occasional, in the artistic gist of the novel “Asan” by V. Makanin, as well as their role in structuring the individual-author’s linguistic picture of the world. It has been proven that the derivation of expressive lexemes is the result of improvisation according to established patterns, and that the formation of occasional substantives, adjectives and verbs involved the main methods of the Russian word derivation. It is noted that in the artistic discourse of V. Makanin, in the substantive word-formation, suffixation plays a leading role, which takes place in the sphere of abstractness and includes such lexical-semantic groups as expressive substantives with the meaning of a person, expressive substantives with the meaning of abstracted action or an abstract feature with connotation, as a rule, negative and/or reduced colloquial connotation. While in the sphere of adjectival and verbal word formation, confixation and prefixation, as the formation of expressiveness, is most productive. The paper considers the phenomenon of semantic word formation, describes the formation of semantic derivatives, including in the field of occasional vocabulary. Expressive derivatives in the artistic discourse of V. Makanin are a bright sign of his individual style, an important means of expressing the world view and outlook of the writer.