Search results

Mouth segmentation is one of the challenging tasks of development in lip reading applications due to illumination, low chromatic contrast and complex mouth appearance. Recently, deep learning methods effectively solved mouth segmentation problems with state-of-the-art performances. This study presents a modified Mobile DeepLabV3 based technique with a comprehensive evaluation based on mouth datasets.

This paper presents a novel approach to mouth segmentation by Mobile DeepLabV3 technique with integrating decode and auxiliary heads. Extensive data augmentation, online hard example mining (OHEM) and transfer learning have been applied. CelebAMask-HQ and the mouth dataset from 15 healthy subjects in the department of rehabilitation medicine, Ramathibodi hospital, are used in validation for mouth segmentation performance.

Extensive data augmentation, OHEM and transfer learning had been performed in this study. This technique achieved better performance on CelebAMask-HQ than existing segmentation techniques with a mean Jaccard similarity coefficient (JSC), mean classification accuracy and mean Dice similarity coefficient (DSC) of 0.8640, 93.34% and 0.9267, respectively. This technique also achieved better performance on the mouth dataset with a mean JSC, mean classification accuracy and mean DSC of 0.8834, 94.87% and 0.9367, respectively. The proposed technique achieved inference time usage per image of 48.12 ms.

The modified Mobile DeepLabV3 technique was developed with extensive data augmentation, OHEM and transfer learning. This technique gained better mouth segmentation performance than existing techniques. This makes it suitable for implementation in further lip-reading applications.

Semantic retrieval and discovery of datasets published as open data remains a challenging task. The datasets inherently originate in the globally distributed web jungle, lacking the luxury of centralized database administration, database schemes, shared attributes, vocabulary, structure and semantics. The existing dataset catalogs provide basic search functionality relying on keyword search in brief, incomplete or misleading textual metadata attached to the datasets. The search results are thus often insufficient. However, there exist many ways of improving the dataset discovery by employing content-based retrieval, machine learning tools, third-party (external) knowledge bases, countless feature extraction methods and description models and so forth.

In this paper, the authors propose a modular framework for rapid experimentation with methods for similarity-based dataset discovery. The framework consists of an extensible catalog of components prepared to form custom pipelines for dataset representation and discovery.

The study proposes several proof-of-concept pipelines including experimental evaluation, which showcase the usage of the framework.

To the best of authors’ knowledge, there is no similar formal framework for experimentation with various similarity methods in the context of dataset discovery. The framework has the ambition to establish a platform for reproducible and comparable research in the area of dataset discovery. The prototype implementation of the framework is available on GitHub.

Classification techniques have been applied to many applications in various fields of sciences. There are several ways of evaluating classification algorithms. The analysis of such metrics and its significance must be interpreted correctly for evaluating different learning algorithms. Most of these measures are scalar metrics and some of them are graphical methods. This paper introduces a detailed overview of the classification assessment measures with the aim of providing the basics of these measures and to show how it works to serve as a comprehensive source for researchers who are interested in this field. This overview starts by highlighting the definition of the confusion matrix in binary and multi-class classification problems. Many classification measures are also explained in details, and the influence of balanced and imbalanced data on each metric is presented. An illustrative example is introduced to show (1) how to calculate these measures in binary and multi-class classification problems, and (2) the robustness of some measures against balanced and imbalanced data. Moreover, some graphical measures such as Receiver operating characteristics (ROC), Precision-Recall, and Detection error trade-off (DET) curves are presented with details. Additionally, in a step-by-step approach, different numerical examples are demonstrated to explain the preprocessing steps of plotting ROC, PR, and DET curves.

Using the methodologies of text mining, this paper examines the implications of US and Chinese policies on bilateral trade. Official speeches by political leaders of the U.S. and China on the issues of trade were collected and analytically examined for US-China gaps in major foreign policies, such as bilateral trade and the Belt and Road Initiative. In this paper, a term frequency-inverse document frequency word cloud, a network similarities index, machine learning-processed latent Dirichlet allocation (LDA), and structural equivalence are applied to examine the meanings of the speeches. The main arguments in this paper are as follows. First, the document similarity between the speeches of Chinese and US leaders appears to be completely different. Also, while the documents from Chinese leaders are considerably similar, the documents from US leaders differ by far. Secondly, LDA topic analysis indicates that China concentrates more on international and collaborative relationships, while the U.S. has more focus on domestic and economic interests. Third, from a word hierarchy analysis, the basic words used by American and Chinese leaders are also completely different. Agriculture, farmers, automobiles, and negotiations are the basic words for American leaders, but for Chinese leaders, the basic words are planning, markets, and education.

The purpose of this paper is to propose a scheme that allows users to interactively explore relations between entities in knowledge bases (KBs). KBs store a wide range of knowledge about real-world entities in a structured form as (subject, predicate, object). Although it is possible to query entities and relations among entities by specifying appropriate query expressions of SPARQL or keyword queries, the structure and the vocabulary are complicated, and it is hard for non-expert users to get the desired information. For this reason, many researchers have proposed faceted search interfaces for KBs. Nevertheless, existing ones are designed for finding entities and are insufficient for finding relations.

To this problem, the authors propose a novel “relation facet” to find relations between entities. To generate it, they applied clustering on predicates for grouping those predicates that are connected to common objects. Having generated clusters of predicates, the authors generated a facet according to the result. Specifically, they proposed to use a couple of clustering algorithms, namely, agglomerative hierarchical clustering (AHC) and CANDECOMP/PARAFAC (CP) tensor decomposition which is one of the tensor decomposition methods.

The authors experimentally show test the performance of clustering methods and found that AHC performs better than tensor decomposition. Besides, the authors conducted a user study and show that their proposed scheme performs better than existing ones in the task of searching relations.

The authors propose a relation-oriented faceted search method for KBs that allows users to explore relations between entities. As far as the authors know, this is the first method to focus on the exploration of relations between entities.

This paper aims to introduce a crowd-based method for theorizing. The purpose is not to achieve a scientific theory. On the contrary, the purpose is to achieve a model that may challenge current scientific theories or lead research in new phenomena.

This paper describes a case study of theorizing by using a crowd-based method. The first section of the paper introduces what do the authors know about crowdsourcing, crowd science and the aggregation of non-expert views. The second section details the case study. The third section analyses the aggregation. Finally, the fourth section elaborates the conclusions, limitations and future research.

This document answers to what extent the crowd-based method produces similar results to theories tested and published by experts.

From a theoretical perspective, this study provides evidence to support the research agenda associated with crowd science. The main limitation of this study is that the crowded research models and the expert research models are compared in terms of the graph. Nevertheless, some academics may argue that theory building is about an academic heritage.

This paper exemplifies how to obtain an expert-level research model by aggregating the views of non-experts.

This study is particularly important for institutions with limited access to costly databases, labs and researchers.

Previous research suggested that a collective of individuals may help to conduct all the stages of a research endeavour. Nevertheless, a formal method for theorizing based on the aggregation of non-expert views does not exist. This paper provides the method and evidence of its practical implications.

The density peak clustering algorithm (DP) is proposed to identify cluster centers by two parameters, i.e. ρ value (local density) and δ value (the distance between a point and another point with a higher ρ value). According to the center-identifying principle of the DP, the potential cluster centers should have a higher ρ value and a higher δ value than other points. However, this principle may limit the DP from identifying some categories with multi-centers or the centers in lower-density regions. In addition, the improper assignment strategy of the DP could cause a wrong assignment result for the non-center points. This paper aims to address the aforementioned issues and improve the clustering performance of the DP.

First, to identify as many potential cluster centers as possible, the authors construct a point-domain by introducing the pinhole imaging strategy to extend the searching range of the potential cluster centers. Second, they design different novel calculation methods for calculating the domain distance, point-domain density and domain similarity. Third, they adopt domain similarity to achieve the domain merging process and optimize the final clustering results.

The experimental results on analyzing 12 synthetic data sets and 12 real-world data sets show that two-stage density peak clustering based on multi-strategy optimization (TMsDP) outperforms the DP and other state-of-the-art algorithms.

The authors propose a novel DP-based clustering method, i.e. TMsDP, and transform the relationship between points into that between domains to ultimately further optimize the clustering performance of the DP.

The purpose of this study is to describe latent classes explaining the innovation logic in the Finnish construction companies. Innovativeness is a driver of competitive performance and vital to the long-term success of any organisation and company.

Using finite mixture structural equation modelling (FMSEM), the authors have classified innovation logic into latent classes. The method analyses and recognises classes for companies that have similar logic in innovation activities based on the collected data.

Through FMSEM analysis, the authors have identified three latent classes that explain the innovation logic in the Finnish construction companies – LC1: the internal innovators; LC2: the non-innovation-oriented introverts; and LC3: the innovation-oriented extroverts. These three latent classes clearly capture the perceptions within the industry as well as the different characteristics and variables.

The presented latent classes explain innovation logic but is limited to analysing Finnish companies. Also, the research is quantitative by nature and does not increase the understanding in the same manner as qualitative research might capture on more specific aspects.

This paper presents starting points for construction industry companies to intensify innovation activities. It may also indicate more fundamental changes for the structure of construction industry organisations, especially by enabling innovation friendly culture.

This study describes innovation logic in Finnish construction companies through three models (LC1–LC3) by using quantitative data analysed with the FMSEM method. The fundamental innovation challenges in the Finnish construction companies are clarified via the identified latent classes.

The research carried out a bibliometric analysis of the literature on environmental sustainability from a demand perspective by analyzing the scientific contributions published in the last twenty years.

A bibliometric analysis was carried out to outline the scientific studies development, identifying the most discussed topics and those that would require future research. In total, 274 articles published between 1999 and 2021 were collected through the Web of Science database and analyzed with the SciMAT software.

By systematizing the literature results, the study revealed a steady growth in the number of publications and in the research areas, highlighting a substantial evolution of the research topic.

The study contribute for conceptual, methodological and thematic development of the topic, systematizing the results of existing studies and providing useful indications for the promotion of sustainable consumer habits.

The study attempts to bridge the gap in current literature by offering a holistic view on the role of consumer behavior in pursuing sustainability goals, identifying both the most treated areas and the emerging ones that can represent opportunities for future research.

The purpose of this paper is to propose a new approach to retrieve similar questions in online health communities to improve the efficiency of health information retrieval and sharing.

This paper proposes a hybrid approach to combining domain knowledge similarity and topic similarity to retrieve similar questions in online health communities. The domain knowledge similarity can evaluate the domain distance between different questions. And the topic similarity measures questions’ relationship base on the extracted latent topics.

The experiment results show that the proposed method outperforms the baseline methods.

This method conquers the problem of word mismatch and considers the named entities included in questions, which most of existing studies did not.