Semantic Space Research Articles

Semantics-Guided Contrastive Network for Zero-Shot Object Detection.

Zero-shot object detection (ZSD), the task that extends conventional detection models to detecting objects from unseen categories, has emerged as a new challenge in computer vision. Most existing approaches tackle the ZSD task with a strict mapping-transfer strategy that may lead to suboptimal ZSD results: 1) the learning process of these models neglects the available semantic information on unseen classes, which can easily bias towards the seen categories; 2) the original visual feature space is not well-structured for the ZSD task due to the lack of discriminative information. To address these issues, we develop a novel Semantics-Guided Contrastive Network for ZSD, named ContrastZSD, a detection framework that first brings contrastive learning mechanism into the realm of zero-shot detection. Particularly, ContrastZSD incorporates two semantics-guided contrastive learning subnets that contrast between region-category and region-region pairs respectively. The pairwise contrastive tasks take advantage of supervision signals derived from both the ground truth label and class similarity information. By performing supervised contrastive learning over those explicit semantic supervision, the model can learn more knowledge about unseen categories to avoid the bias problem to seen concepts, while optimizing the visual data structure to be more discriminative for better visual-semantic alignment. Extensive experiments are conducted on two popular benchmarks for ZSD, i.e., PASCAL VOC and MS COCO. Results show that our method outperforms the previous state-of-the-art on both ZSD and generalized ZSD tasks.

The effectiveness of connotation to produce spatial transformation in the Iraqi theatrical performance

Place is an active and essential element in the construction and production of dramatic action because of its connection to the event. It is the place that contains the event and the action, and thus it has a psychological and semantic impact on thought and meaning. On the other hand, the place is connected to time and does not protect it from semantic manifestations and transformations. Thus, the place takes up its semantic space within the system. The theatrical presentation is an active and influential part in the structure of the event. The theatrical presentation witnessed spatial changes and variables that intertwine in the fabric of the intellectual structure to transform into connotations and symbols that carry and investigate meaning. The place was not fixed and rigid, but rather a dynamic, active variable. According to this, the research headed to monitor the effectiveness of connotation in producing transformation. Spatial within the theatrical performance.

Fragrance lexicon for analysis of consumer-generated perfume reviews in Russian and English

In recent years, there has been a rise in research on sensorium in various academic disciplines. Olfaction is recognized as a sense that is most closely linked to cognition, memory and emotion. Due to this unique feature, studies on various aspects of human olfaction are steadily gaining prominence in the humanities and social sciences. In order to understand how the olfactory modality is marked, several taxonomies and semantic spaces of olfactory terms have been developed. However, the focus has been on the general olfaction lexicon and there is a lack of systematic and comprehensive lexicons for fragrant smells. This article addresses this gap. It adopts a multilingual perspective and describes the process of developing a fragrance lexicon in two languages, Russian and English. A fragrance lexicon refers to a list of words that people might use to describe a perfume. The steps in the lexicon development included•sourcing the lexical items in the two languages•translating and cleaning the word lists•revising and refining the lexiconThe fragrance lexicon presented in this article can be used to aid linguistic analyses of naturally occurring communications about perfumes, such as computational analyses of consumer-generated perfume reviews.

Using word embeddings for immigrant and refugee stereotype quantification in a diachronic and multilingual setting

Word embeddings are efficient machine-learning-based representations of human language used in many Natural Language Processing tasks nowadays. Due to their ability to learn underlying word association patterns present in large volumes of data, it is possible to observe various sociolinguistic phenomena in the embedding semantic space, such as social stereotypes. The use of stereotypical framing in discourse can be detrimental and induce misconceptions about certain groups, such as immigrants and refugees, especially when used by media and politicians in public discourse. In this paper, we use word embeddings to investigate immigrant and refugee stereotypes in a multilingual and diachronic setting. We analyze the Danish, Dutch, English, and Spanish portions of four different multilingual corpora of political discourse, covering the 1997–2018 period. Then, we measure the effect of sociopolitical variables such as the number of offences committed and the size of the refugee and immigrant groups in the host country over our measurements of stereotypical association using the Bayesian multilevel framework. Our results indicate the presence of stereotypical associations towards both immigrants and refugees for all 4 languages, and that the immigrants are overall more strongly associated with the stereotypical frames than refugees.

Structure-aware contrastive hashing for unsupervised cross-modal retrieval

Cross-modal hashing has attracted a lot of attention and achieved remarkable success in large-scale cross-media similarity retrieval applications because of its superior computational efficiency and low storage overhead. However, constructing similarity relationship among samples in cross-modal unsupervised hashing is challenging because of the lack of manual annotation. Most existing unsupervised methods directly use the representations extracted from the backbone of their respective modality to construct instance similarity matrices, leading to inaccurate similarity matrices and resulting in suboptimal hash codes. To address this issue, a novel unsupervised hashing model, named Structure-aware Contrastive Hashing for Unsupervised Cross-modal Retrieval (SACH), is proposed in this paper. Specifically, we concurrently employ both high-dimensional representations and discriminative representations learned by the network to construct a more informative semantic correlative matrix across modalities. Moreover, we design a multimodal structure-aware alignment network to minimize heterogeneous gap in the high-order semantic space of each modality, effectively reducing disparities within heterogeneous data sources and enhancing the consistency of semantic information across modalities. Extensive experimental results on two widely utilized datasets demonstrate the superiority of our proposed SACH method in cross-modal retrieval tasks over existing state-of-the-art methods.

Trustworthy multi-view clustering via alternating generative adversarial representation learning and fusion

Multi-view clustering (MVC) has attached extensive attention as it provides an effective approach to deal with the unlabeled data in real-world applications. To enjoy the strong feature extraction capacity of deep learning, traditional shallow MVC methods are further extended to deep version. Though achieve superiorities in many fields, most existing deep MVC models are still limited by their lacking assurance of the trustworthiness and robustness. This drawback also leads to the possible poor clustering performance. To tackle this drawback, we propose an alternating generative adversarial learning strategy based network, termed as Alternating Generative Adversarial Representation Learning (AGARL) network for multi-view clustering, aiming to drive different views to fall into the same semantic space for the further alignment and information fusion. Various from the traditional generative adversarial learning strategy, the proposed alternating adversarial representation learning strategy successes in further aligning the distributions of the view-specific representations and exploring the cross-view consistency credited by its setting each view as the reference alternately. Then a consensus representative latent representation can be learned through an attention-like sub-network for all views. Finally, a self-supervised clustering module is employed in the proposed network to guide the learning of cluster assignment. In addition, comprehensive experiments on numerous benchmark MVC datasets have been performed to demonstrate the effectiveness of the proposed network compared to other state-of-the-art methods.

Representational similarity learning reveals a graded multidimensional semantic space in the human anterior temporal cortex

Abstract Neurocognitive models of semantic memory have proposed that the ventral anterior temporal lobes (vATLs) encode a graded and multidimensional semantic space—yet neuroimaging studies seeking brain regions that encode semantic structure rarely identify these areas. In simulations, we show that this discrepancy may arise from a crucial mismatch between theory and analysis approach. Utilizing an analysis recently formulated to investigate graded multidimensional representations, representational similarity learning (RSL), we decoded semantic structure from ECoG data collected from the vATL cortical surface while participants named line drawings of common items. The results reveal a graded, multidimensional semantic space encoded in neural activity across the vATL, which evolves over time and simultaneously expresses both broad and finer-grained semantic structure among animate and inanimate concepts. The work resolves the apparent discrepancy within the semantic cognition literature and, more importantly, suggests a new approach to discovering representational structure in neural data more generally.

Applying kansei engineering and text mining to design fresh food logistics services

ABSTRACT The study focuses on identifying the key drivers that can meet consumers’ perceptual demands for fresh food logistics services and how e-retailers balance changing demands. In this light, a service quality improvement approach integrating Kansei engineering and text mining is proposed to span the property and semantic space. The data from the Chinese e-commerce platform is used as 288 service scenarios. Meanwhile, GeoDetector and Kano model are used to build models and find priorities of services. The results show that consumers’ demands are fine-grained and hierarchical. Fresh, fast, affordable, reliable and satisfied demands are the rational. Convenient, thoughtful, pleasant, persistent and surprised demands are the perceptual. Most logistics attributes have a ‘1 + 1 > 2’ effect on consumers’ perceptual demands during the emergency period, and exposure safety and logistics persistence are most significantly affected, having a ‘weak + weak = strong’ effect. Additionally, must-be, one-dimensional, and attractive quality based on the kano model provide more targeted strategies to improve the input-output utility.

GNNCL: A Graph Neural Network Recommendation Model Based on Contrastive Learning

In the field of recommendation algorithms, the representation learning for users and items has evolved from using single IDs or historical interactions to utilizing higher-order neighbors. This can be achieved by modeling the user–item interaction graph to capture user preferences for items. Despite the promising results achieved by these algorithms, they still suffer from the issue of data sparsity. In order to mitigate the impact of data sparsity, contrastive learning has been adopted in graph collaborative filtering to enhance performance. However, current recommendation algorithms using contrastive learning yield uneven representations after data augmentation and do not consider the potential relationships among users (or items). To address these challenges, we propose a graph neural network-based recommendation model that integrates contrastive learning (GNNCL). This model combines data augmentation with added noise and the exploration of semantic neighbors for nodes. For the structural neighbors on the interaction graph, we introduce a novel and straightforward contrastive learning approach, abandoning previous graph augmentation methods, and introducing uniform noise into the embedding space to create contrastive views. To unearth potential semantic neighbor relationships in the semantic space, we assume that users with similar representations possess semantic neighbor relationships and merge these semantic neighbors into the prototype contrastive learning. We utilize a clustering algorithm to obtain prototypes for users and items and employ the EM algorithm for prototype contrastive learning. Experimental results validate the effectiveness of our approach. Particularly, on the Yelp2018 and Amazon-book datasets, our method exhibits significant performance improvements compared to basic graph collaborative filtering models.

Overlapping representations of observed actions and action-related features.

The lateral occipitotemporal cortex (LOTC) has been shown to capture the representational structure of a smaller range of actions. In the current study, we carried out an fMRI experiment in which we presented human participants with images depicting 100 different actions and used representational similarity analysis (RSA) to determine which brain regions capture the semantic action space established using judgments of action similarity. Moreover, to determine the contribution of a wide range of action-related features to the neural representation of the semantic action space we constructed an action feature model on the basis of ratings of 44 different features. We found that the semantic action space model and the action feature model are best captured by overlapping activation patterns in bilateral LOTC and ventral occipitotemporal cortex (VOTC). An RSA on eight dimensions resulting from principal component analysis carried out on the action feature model revealed partly overlapping representations within bilateral LOTC, VOTC, and the parietal lobe. Our results suggest spatially overlapping representations of the semantic action space of a wide range of actions and the corresponding action-related features. Together, our results add to our understanding of the kind of representations along the LOTC that support action understanding.

Heuristics and semantic spaces for the analysis of students’ work in mathematical problem solving

In this research, our objective is to characterize the problem-solving procedures of primary and lower secondary students when they solve problems in real class conditions. To do so, we rely first on the concept of heuristics. As this term is very polysemic, we exploit the definition proposed by Rott (2014) to develop a coding manual and thus analyze students’ procedures. Then, we interpret the results of these analyses in a qualitative way by mobilizing the concept of semantic space (Poitrenaud, 1998). This detailed analysis of students’ procedures is made possible by collecting audiovisual data as close as possible to the students’ work using an action camera mounted on the students’ heads. We thus succeed in highlighting three different investigation profiles that we have named explorer, butterfly, and prospector. Our first results tend to show a correlation with these profiles and the success in problem-solving, yet this would need more investigation.

Identification of gene biomarkers for brain diseases via multi-network topological semantics extraction and graph convolutional network

BackgroundBrain diseases pose a significant threat to human health, and various network-based methods have been proposed for identifying gene biomarkers associated with these diseases. However, the brain is a complex system, and extracting topological semantics from different brain networks is necessary yet challenging to identify pathogenic genes for brain diseases.ResultsIn this study, we present a multi-network representation learning framework called M-GBBD for the identification of gene biomarker in brain diseases. Specifically, we collected multi-omics data to construct eleven networks from different perspectives. M-GBBD extracts the spatial distributions of features from these networks and iteratively optimizes them using Kullback–Leibler divergence to fuse the networks into a common semantic space that represents the gene network for the brain. Subsequently, a graph consisting of both gene and large-scale disease proximity networks learns representations through graph convolution techniques and predicts whether a gene is associated which brain diseases while providing associated scores. Experimental results demonstrate that M-GBBD outperforms several baseline methods. Furthermore, our analysis supported by bioinformatics revealed CAMP as a significantly associated gene with Alzheimer's disease identified by M-GBBD.ConclusionCollectively, M-GBBD provides valuable insights into identifying gene biomarkers for brain diseases and serves as a promising framework for brain networks representation learning.

Measuring and Mitigating Gender Bias in Legal Contextualized Language Models

Transformer-based contextualized language models constitute the state-of-the-art in several natural language processing (NLP) tasks and applications. Despite their utility, contextualized models can contain human-like social biases, as their training corpora generally consist of human-generated text. Evaluating and removing social biases in NLP models has been a major research endeavor. In parallel, NLP approaches in the legal domain, namely, legal NLP or computational law, have also been increasing. Eliminating unwanted bias in legal NLP is crucial, since the law has the utmost importance and effect on people. In this work, we focus on the gender bias encoded in BERT-based models. We propose a new template-based bias measurement method with a new bias evaluation corpus using crime words from the FBI database. This method quantifies the gender bias present in BERT-based models for legal applications. Furthermore, we propose a new fine-tuning-based debiasing method using the European Court of Human Rights (ECtHR) corpus to debias legal pre-trained models. We test the debiased models’ language understanding performance on the LexGLUE benchmark to confirm that the underlying semantic vector space is not perturbed during the debiasing process. Finally, we propose a bias penalty for the performance scores to emphasize the effect of gender bias on model performance.

Multi-level video captioning method based on semantic space

Multi-level video captioning method based on semantic space

CycleSeg: Simultaneous synthetic CT generation and unsupervised segmentation for MR-only radiotherapy treatment planning of prostate cancer.

MR-only radiotherapy treatment planning is an attractive alternative to conventional workflow, reducing scan time and ionizing radiation. It is crucial to derive the electron density map or synthetic CT (sCT) from MR data to perform dose calculations to enable MR-only treatment planning. Automatic segmentation of relevant organs in MR images can accelerate the process by preventing the time-consuming manual contouring step. However, the segmentation label is available only for CT data in many cases. We propose CycleSeg, a unified framework that generates sCT and corresponding segmentation from MR images without access to MR segmentation labels METHODS: CycleSeg utilizes the CycleGAN formulation to perform unpaired synthesis of sCT and image alignment. To enable MR (sCT) segmentation, CycleSeg incorporates unsupervised domain adaptation by using a pseudo-labeling approach with feature alignment in semantic segmentation space. In contrast to previous approaches that perform segmentation on MR data, CycleSeg could perform segmentation on both MR and sCT. Experiments were performed with data from prostate cancer patients, with 78/7/10 subjects in the training/validation/test sets, respectively. CycleSeg showed the best sCT generation results, with the lowest mean absolute error of 102.2 and the lowest Fréchet inception distance of 13.0. CycleSeg also performed best on MR segmentation, with the highest average dice score of 81.0 and 81.1 for MR and sCT segmentation, respectively. Ablation experiments confirmed the contribution of the proposed components of CycleSeg. CycleSeg effectively synthesized CT and performed segmentation on MR images of prostate cancer patients. Thus, CycleSeg has the potential to expedite MR-only radiotherapy treatment planning, reducing the prescribed scans and manual segmentation effort, and increasing throughput.

Unisolated locality, or A cultural universal in action

The author assumes that the former idea of the cultural universal implied the danger of unification for a national and cultural identity associated with something conventional, virtual, essentially spurious, and thus divorced from the national-literary discourse. The exaggerated identification of tradition and the universal blocked a further extension of the semantic space of literature. The article considers the mutually complementary nature of tradition, on the one hand, and novelty, singularity, and a consolidating supra-ethnic idea, on the other: to preserve itself, individuality needs communicativeness, whereas ‘the force of universality,’ according to Hegel, ‘contains particularity.’ The unique and individual qualities of each national literature suggest unisolated locality and, therefore, the universality of the particular. In the specific literary work (A. Kim’s prose) the article traces the author’s predisposition for a holistic worldview that combines a national mode and a universalization paradigm.

Exploiting Cross-Modal Prediction and Relation Consistency for Semisupervised Image Captioning.

The task of image captioning aims to generate captions directly from images via the automatically learned cross-modal generator. To build a well-performing generator, existing approaches usually need a large number of described images (i.e., supervised image-sentence pairs), requiring a huge effects on manual labeling. However, in real-world applications, a more general scenario is that we only have limited amount of described images and a large number of undescribed images. Therefore, a resulting challenge is how to effectively combine the undescribed images into the learning of cross-modal generator (i.e., semisupervised image captioning). To solve this problem, we propose a novel image captioning method by exploiting the cross-modal prediction and relation consistency (CPRC), which aims to utilize the raw image input to constrain the generated sentence in the semantic space. In detail, considering that the heterogeneous gap between modalities always leads to the supervision difficulty while using the global embedding directly, CPRC turns to transform both the raw image and corresponding generated sentence into the shared semantic space, and measure the generated sentence from two aspects: 1) prediction consistency: CPRC utilizes the prediction of raw image as soft label to distill useful supervision for the generated sentence, rather than employing the traditional pseudo labeling and 2) relation consistency: CPRC develops a novel relation consistency between augmented images and corresponding generated sentences to retain the important relational knowledge. In result, CPRC supervises the generated sentence from both the informativeness and representativeness perspectives, and can reasonably use the undescribed images to learn a more effective generator under the semisupervised scenario. The experiments show that our method outperforms state-of-the-art comparison methods on the MS-COCO "Karpathy" offline test split under complex nonparallel scenarios, for example, CPRC achieves at least 6% improvements on the CIDEr-D score.

Realization of the Diachronic Emergence of the Text Space (Using the Example of a Creolized Text)

The subject of this study is the specifics of the implementation of the diachronic emergence of the text space on the example of hypertext. From the point of view of psycholinguistics and linguosynergetics, scientific interest is the result of understanding the text space. In this regard, the relevance of the research topic lies in the ambiguity of the movement of the text space, manifested in the increment of meaning as a result of prompting an individual to speech activity in order to project emotionally significant information. Such a process in linguosynergetic science is commonly called emergence. This phenomenon is the result of the perception of a creolized text, and also structures fragments of semantic and associative semantic fields. The article presents the results of a directed psycholinguistic experiment that demonstrated the process of increment of meaning formed as a result of the perception of a complex text formation. The methodology of the organization and conduct of experimental work is described. The scientific novelty of the research is reflected in the use of a new approach to the study of the specifics of the development of meaning. The directed psycholinguistic experiment made it possible to determine the role of emergence in the formation of the dynamics of semantic text space. Along with this, the possibilities of the phenomenon under study are highlighted, which are revealed in the formation of new systems, which is a demonstration of the movement of the text space. As a result, two groups of emergence components have been identified: the first is represented by comments developing the main idea, distributed by topics; the second is contradictions, which are logically contradictory statements that do not correspond to the topic of the stimulus material. In the course of the implementation of the information flow of the text, new meanings, thematic ramifications arise, the potential possibility of anticipating which is significant for diachronic emergence.

The specifics and functions of the concepts of LIFE and DEATH in V. Y. Bryusov's poetry

The object of the study are poems representing such thematic layers as love lyrics; urban lyrics, including poems dedicated to Venice; the theme of death; the theme of Vikings; great figures of the past; apocalyptic themes. The paper actualizes the importance of studying Bryusov's poetry in the aspect of cognitive linguistics, examines the essence of the concepts. The place of the concepts of LIFE and DEATH in the hierarchical structure of the conceptual sphere is analyzed, their meanings are revealed based on the lexicographic material of dictionaries, representing, along with the primary meanings, various associative lexical units, including mythologems that arose on the basis of cognitive metaphorization, which, being representatives of concepts, fill slots, form a frame structure, while the frame qualifies as the core of the concept. The analysis of poems allowing to recreate the linguistic and artistic worldview of the poet, the keys to understanding which are the basic concepts of LIFE and DEATH, and to determine their specificity and role, is carried out. The following methods were used in the research process: the method of discursive analysis, which allowed us to study the interacting denotative and conceptual structures that organize the semantic space of each lyrical work under consideration in the context of Bryusov's poetry; the cultural and historical method, which contributed to the identification of connotative meanings representing the concepts of LIFE and DEATH; the hermeneutic method, which made it possible to analyze the poetics of poems. As a result of the study, the authors came to the following conclusions. Organizing a system of poetic creativity, vividly embodying the main motive, the motive of duality, the concepts of LIFE and DEATH undergo semantic contamination based on the intersection of their frame networks and form a single bipolar concept of LIFE/DEATH. An integral phenomenon, Bryusov's concept of LIFE/DEATH is multifunctional: as a figurative means, this structural unit creates the poet's idiosyncrasy; plays a text-forming role; organizes the main motives that allow Bryusov to carry out a discourse relevant to him and his contemporaries about the problem of the meaning of life and its eternity. The scientific novelty lies in the fact that the work of V. Y. Bryusov, from the point of view of the conceptual sphere and conceptualization, has not been sufficiently studied. For the first time, the comprehension of V.Y. Bryusov's linguistic picture of the world was realized through the concepts of LIFE and DEATH, which are fundamental in his work.

Exploring Semantic Redundancy using Backdoor Triggers: A Complementary Insight into the Challenges facing DNN-based Software Vulnerability Detection

To detect software vulnerabilities with better performance, deep neural networks (DNNs) have received extensive attention recently. However, these vulnerability detection DNN models trained with code representations are vulnerable to specific perturbations on code representations. This motivates us to rethink the bane of software vulnerability detection and find function-agnostic features during code representation which we name as semantic redundant features. This paper first identifies a tight correlation between function-agnostic triggers and semantic redundant feature space (where the redundant features reside) in these DNN models. For correlation identification, we propose a novel Backdoor-based Semantic Redundancy Exploration (BSemRE) framework. In BSemRE, the sensitivity of the trained models to function-agnostic triggers is observed to verify the existence of semantic redundancy in various code representations. Specifically, acting as the typical manifestations of semantic redundancy, naming conventions, ternary operators and identically-true conditions are exploited to generate function-agnostic triggers. Extensive comparative experiments on 1613823 samples of 8 representative vulnerability datasets and state-of-the-art code representation techniques and vulnerability detection models demonstrate that the existence of semantic redundancy determines the upper trustworthiness limit of DNN-based software vulnerability detection. To the best of our knowledge, this is the first work exploring the bane of software vulnerability detection using backdoor triggers.