DeepVisInterests : Deep data analysis for topics of interests prediction

Deep data analysis for latent information prediction has been an important research area. Many of the existing solutions have used the textual data and have obtained an accurate results for predicting users' interests and other latent attributes. However, little attention has been paid to visual data that is becoming increasingly popular in recent times. In this paper, we addresses the problem of discovering the attributed interest and of analyzing the performance of the automatic prediction using a comparison with the self assessed topics of interest (topics of interest provided by the user in a proposed questionnaire) based on data analysis techniques applied on the users visual data. We analyze the content of each user's images to aggregate the image-level users' interests distribution in order to obtain the user-level users' interest distribution. To do this, we employ the pretrained ImageNet convolutional neural networks architectures for the feature extraction step and to construct the ontology, as the users' interests model, in order to learn the semantic representation for the popular topics of interests defined by social networks (e.g., Facebook). Our experimental studies show that this analysis, on the most relevant features, enhances the performance of the prediction framework. In order to improve our framework's robustness and generalization with unknown users' profiles, we propose a novel database evaluation. Our proposed framework provided promising results which are competitive to state-of-the-art techniques with an accuracy of 0.80.

DeepVisInterests : Deep data analysis for topics of interests prediction

Deep data analysis for latent information prediction has been an important research area. Many of the existing solutions have used the textual data and have obtained an accurate results for predicting users' interests and other latent attributes. However, little attention has been paid to visual data that is becoming increasingly popular in recent times. In this paper, we addresses the problem of discovering the attributed interest and of analyzing the performance of the automatic prediction using a comparison with the self assessed topics of interest (topics of interest provided by the user in a proposed questionnaire) based on data analysis techniques applied on the users visual data. We analyze the content of each user's images to aggregate the image-level users' interests distribution in order to obtain the user-level users' interest distribution. To do this, we employ the pretrained ImageNet convolutional neural networks architectures for the feature extraction step and to construct the ontology, as the users' interests model, in order to learn the semantic representation for the popular topics of interests defined by social networks (e.g., Facebook). Our experimental studies show that this analysis, on the most relevant features, enhances the performance of the prediction framework. In order to improve our framework's robustness and generalization with unknown users' profiles, we propose a novel database evaluation. Our proposed framework provided promising results which are competitive to state-of-the-art techniques with an accuracy of 0.80.

FORECASTING THE INTENSITY OF SOLAR RADIATION BASED ON ARTIFICIAL NEURAL NETWORKS

The paper considers short-term forecasting of the intensity of solar radiation in the city of Odessa based on an artificial neural network. The artificial neural network was trained on the experimental data of the ground weather station (Davis 6162EU), which is installed on the roof of the educational building of the Odessa National Polytechnic University. Modeling, validation, and testing of experimental data were performed using the MATLAB software package, namely Neural Network Toolbox. The Levenberg-Markwatt model is used in this work. The analyzed data set was divided into proportions of 70%, 15%, 15% for neural network training, its validation, and testing, respectively. The results which the trained neural network gave during forecasting within the framework of the database and outside it are given. The deviation between real and forecast data is analyzed. The root-mean-square error on December 26, 2016 was 13.03 W / m2, and on December 27, 2016 - 9.44 W / m2 when forecasting outside the database. Evaluation of the accuracy of an artificial neural network has shown its effectiveness in predicting the intensity of solar radiation. To predict parameters based on artificial neural networks, experimental data that describe a real system are needed. Artificial neural networks, like other approximation methods, have both advantages and disadvantages.

Histoplasma capsulatum Glycans From Distinct Genotypes Share Structural and Serological Similarities to Cryptococcus neoformans Glucuronoxylomannan

The cell wall is a ubiquitous structure in the fungal kingdom, with some features varying depending on the species. Additional external structures can be present, such as the capsule of Cryptococcus neoformans (Cn), its major virulence factor, mainly composed of glucuronoxylomannan (GXM), with anti-phagocytic and anti-inflammatory properties. The literature shows that other cryptococcal species and even more evolutionarily distant species, such as the Trichosporon asahii, T. mucoides, and Paracoccidioides brasiliensis can produce GXM-like polysaccharides displaying serological reactivity to GXM-specific monoclonal antibodies (mAbs), and these complex polysaccharides have similar composition and anti-phagocytic properties to cryptococcal GXM. Previously, we demonstrated that the fungus Histoplasma capsulatum (Hc) incorporates, surface/secreted GXM of Cn and the surface accumulation of the polysaccharide enhances Hc virulence in vitro and in vivo. In this work, we characterized the ability of Hc to produce cellular-attached (C-gly-Hc) and secreted (E-gly) glycans with reactivity to GXM mAbs. These C-gly-Hc are readily incorporated on the surface of acapsular Cn cap59; however, in contrast to Cn GXM, C-gly-Hc had no xylose and glucuronic acid in its composition. Mapping of recognized Cn GXM synthesis/export proteins confirmed the presence of orthologs in the Hc database. Evaluation of C-gly and E-gly of Hc from strains of distinct monophyletic clades showed serological reactivity to GXM mAbs, despite slight differences in their molecular dimensions. These C-gly-Hc and E-gly-Hc also reacted with sera of cryptococcosis patients. In turn, sera from histoplasmosis patients recognized Cn glycans, suggesting immunogenicity and the presence of cross-reacting antibodies. Additionally, C-gly-Hc and E-gly-Hc coated Cn cap59 were more resistant to phagocytosis and macrophage killing. C-gly-Hc and E-gly-Hc coated Cn cap59 were also able to kill larvae of Galleria mellonella. These GXM-like Hc glycans, as well as those produced by other pathogenic fungi, may also be important during host-pathogen interactions, and factors associated with their regulation are potentially important targets for the management of histoplasmosis.

Simultaneous prediction of valence / arousal and emotion categories and its application in an HRC scenario

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.

The 2020 IDS Release of the Antwerp COR*-Database. Evaluation, Development and Transformation of a Pre-Existing Database

The Antwerp COR*-IDS database 2020 is a transformed and harmonized historical demographic database in a cross-nationally comparable format designed to be open and easy to use for international researchers. The database is constructed from the 2010 release of the Antwerp COR*-historical demographic database, which was created using a letter sample of the whole district of Antwerp (Flanders, Belgium). It has a total sample size of +/- 33,000 residents of Antwerp. The sample spans nearly seven decades. The data is collected from historical records: including population registers and vital registration records covering births, marriages, in/external migrations and deaths. The database covers up to three linked generations (in some cases more), and contains micro-data on individual level life courses, and relationships deriving from addressbased household composition methods. An important characteristic is the sample's large migrant population, including the timings of their demographic events and living arrangements, whilst resident in the district of Antwerp. In addition, the sample also contains a large array of occupational level information. This paper presents the processes, methodologies and documentation regarding the evaluation and development of a pre-existing historical database. This includes the systematic evaluation of the original samples, methodologies for address based reconstructing of households, and the geocoding of a historical database which took place during the current development of this new version of the database.

Development of a Database Quality Assurance Score (dQAS) to Validate the Nigeria AIDS Indicator and Impact Survey (NAIIS) Database (Preprint)

BACKGROUND In 2018, Nigeria implemented the world’s largest HIV survey “The Nigeria AIDS Indicator and Impact survey (NAIIS)”, with the overarching goal of obtaining more reliable metrics regarding the national scope of HIV epidemic control in Nigeria. OBJECTIVE To (1) describe the processes involved in the development of a new database evaluation tool (“Database Quality Assurance Score [dQAS]) (2) Assess the application of the dQAS in the evaluation and validation of the NAIIS database. METHODS With the assistance of expert review panelists, the dQAS tool was created using an online Delphi (e-Delphi) methodology. Thematic categories were developed to form superordinate categories that grouped themes together. Subordinate categories were then created that decomposed themes down for more specificity. A validation score was employed to assess the technical performance of the NAIIS database. This culminated in the development of the dQAS. RESULTS The finalized dQAS tool was composed of 34 items with a total score of 81.The tool has two sections; a validation item section - which contains 5 sub-sections; and a quality assessment score section, which assigns a score of “1” for “Yes” to indicate that the performance measure item was present and “0” for “No” to indicate that the measure was absent. There were also additional scaling scores ranging from “0” to a maximum of “4” depending on the measure. The NAIIS database achieved 78 out of the maximum total score of 81, yielding an overall technical performance score of 96.3%, which placed it in the highest category denoted as “Exceptional”. CONCLUSIONS This study showed the feasibility of remote internet-based collaboration for the development of dQAS - a tool to assess the validity of a locally-created database infrastructure for a developing setting. Using dQAS, the NAIIS database was found to be valid, reliable and a valuable source of data for future population-based HIV-related studies.

Linking Web Resources in Web of Data to Encyclopedic Knowledge Base

This paper introduces Wikipedia as an extensive knowledge base which provides additional information about a great number of web resources in the semantic web, and shows how RDF web resources in the web of data can be linked to this encyclopedia. Given an input web resource, the designed system identifies the topic of the web resource and links it to the corresponding Wikipedia article. To perform this task, we use the core labeling properties in web of data to specify the candidate Wikipedia articles for a web resource. Finally, a knowledge based approach is used to identify the most appropriate article in Wikipedia database. Evaluation of the system shows the high performance of the designed system.