Annotation Model Research Articles

Large-scale image annotation with image–text hybrid learning models

Managing large-scale image data becomes an important research issue due to the considerably increasing of digital images of late years. For retrieving images by semantic keywords effectively, annotating appropriate concept labels to the corresponding images in advance is required. Many image annotation approaches and models have been proposed in recent years. However, most of the models only focus on analyzing one of the relationships between image visual features and concept texts. In this paper, all the possible relationships of crossing image and text including image-to-text, text-to-text, and image-to-image are considered and discussed. A set of hybrid learning models based on the proposed cross image---text annotation framework are developed and implemented by means of image classifiers, similarity image matching and association mining of image labels. The goal of experiments is to investigate the performance of the cross image---text framework by evaluating the effectiveness of different annotation models including individual models, bi-hybrid models and the all-hybrid model. The results show that not all-hybrid models can improve the accuracy of image annotation. In general, the hybrid models combining the relationships with both images and text boost the effectiveness of annotation.

Annotation-retrieval reinforcement by visual cognition modeling on manifold

While content-based image annotation and retrieval have been active research topics over the past decade, their correlation is not well exploited until recently. We argue that offline annotation and online retrieval models should be regarded as a unified long-term learner to reinforce each other. Based on this viewpoint, this paper presents an annotation-retrieval reinforcement framework, in which a dual-structured learning model is presented: the keyword-image association is learned by semantic manifold modeling while the semantic correlation between keywords is learned by Image-Word Net modeling. In keyword-image association modeling level, to effectively model keyword-image manifold, we present a manifold co-training algorithm to address the sample insufficiency problem. In manifold-based image annotation, we view annotation process as semantic feature reduction in keyword space, based on which a Biased Fisher Discriminant Analysis (BFDA) algorithm is presented for Eigen feature (keyword) selection. In semantic correlation modeling level, a novel Image-Word Net is learned from annotation training set and users' retrieval log for (1) irrelevant annotated keyword pruning; (2) semantic-level retrieval enhancement. In retrieval, our framework can effectively reveal user target by improving traditional content-based relevance feedback to linguistic-level interaction using annotation information, based on which BFDA is adopted for keyword selection. Finally, user interaction logs are adopted for manifold model updating and annotation refinement. As presented in our experiments, proposed method outperforms state-of-the-art annotation and retrieval algorithms in COREL Image databases with over 10,000 general-purpose images.

Enhancing the Enrichment of Pharmacophore-Based Target Prediction for the Polypharmacological Profiles of Drugs.

PharmMapper is a web server for drug target identification by reversed pharmacophore matching the query compound against an annotated pharmacophore model database, which provides a computational polypharmacology prediction approach for drug repurposing and side effect risk evaluation. But due to the inherent nondiscriminative feature of the simple fit scores used for prediction results ranking, the signal/noise ratio of the prediction results is high, posing a challenge for predictive reliability. In this paper, we improved the predictive accuracy of PharmMapper by generating a ligand-target pairwise fit score matrix from profiling all the annotated pharmacophore models against corresponding ligands in the original complex structures that were used to extract these pharmacophore models. The matrix reflects the noise baseline of fit score distribution of the background database, thus enabling estimation of the probability of finding a given target randomly with the calculated ligand-pharmacophore fit score. Two retrospective tests were performed which confirmed that the probability-based ranking score outperformed the simple fit score in terms of identification of both known drug targets and adverse drug reaction related off-targets.

Synchronous communication in PLM environments using annotated CAD models

The connection of resources, data, and knowledge through communication technology plays a vital role in current collaborative design methodologies and Product Lifecycle Management (PLM) systems, as these elements act as channels for information and meaning. Despite significant advances in the area of PLM, most communication tools are used as separate services that are disconnected from existing development environments. Consequently, during a communication session, the specific elements being discussed are usually not linked to the context of the discussion, which may result in important information getting lost or becoming difficult to access. In this paper, we present a method to add synchronous communication functionality to a PLM system based on annotated information embedded in the CAD model. This approach provides users a communication channel that is built directly into the CAD interface and is valuable when individuals need to be contacted regarding the annotated aspects of a CAD model. We present the architecture of a new system and its integration with existing PLM systems, and describe the implementation details of an annotation-based video conferencing module for a commercial CAD application.

Extending Paralldroid with object oriented annotations

Paralldroid unifies the different programming models of Android.The automatic generation of code for the different programming models in Android.We extend the OpenMP annotation model to the object oriented programming paradigm.Paralldroid adapts techniques used in supercomputing systems to handheld systems. The popularity of the handheld systems ( smartphones, tablets , ...) and their computational capability open new challenges in terms of the efficient use of such devices. The heterogeneity of these SoCs and MPSoCs demands very specific knowledge of the devices, involving a very high learning curve for the programmers. To ease the development task we build Paralldroid, a framework oriented to general purpose programmers for mobile devices. Paralldroid unifies the Android programming models and allows for the automatic generation of parallel code. Sections of code to be optimized in a Java program can be annotated using Paralldroid annotations. Paralldroid automatically generates the native C or Renderscript code required to take advantage of the underlying parallel platform (GPU included). The code generated by Paralldroid offers a good performance with a very low cost of development, contributing to increased productivity when developing efficient code.

Leveraging Comparative Genomics to Identify and Functionally Characterize Genes Associated with Sperm Phenotypes in Python bivittatus (Burmese Python).

Comparative genomics approaches provide a means of leveraging functional genomics information from a highly annotated model organism's genome (such as the mouse genome) in order to make physiological inferences about the role of genes and proteins in a less characterized organism's genome (such as the Burmese python). We employed a comparative genomics approach to produce the functional annotation of Python bivittatus genes encoding proteins associated with sperm phenotypes. We identify 129 gene-phenotype relationships in the python which are implicated in 10 specific sperm phenotypes. Results obtained through our systematic analysis identified subsets of python genes exhibiting associations with gene ontology annotation terms. Functional annotation data was represented in a semantic scatter plot. Together, these newly annotated Python bivittatus genome resources provide a high resolution framework from which the biology relating to reptile spermatogenesis, fertility, and reproduction can be further investigated. Applications of our research include (1) production of genetic diagnostics for assessing fertility in domestic and wild reptiles; (2) enhanced assisted reproduction technology for endangered and captive reptiles; and (3) novel molecular targets for biotechnology-based approaches aimed at reducing fertility and reproduction of invasive reptiles. Additional enhancements to reptile genomic resources will further enhance their value.

A discriminative and sparse topic model for image classification and annotation

Image classification is to assign a category of an image and image annotation is to describe individual components of an image by using some annotation terms. These two learning tasks are strongly related. The main contribution of this paper is to propose a new discriminative and sparse topic model (DSTM) for image classification and annotation by combining visual, annotation and label information from a set of training images. The essential features of DSTM different from existing approaches are that (i) the label information is enforced in the generation of both visual words and annotation terms such that each generative latent topic corresponds to a category; (ii) the zero-mean Laplace distribution is employed to give a sparse representation of images in visual words and annotation terms such that relevant words and terms are associated with latent topics. Experimental results demonstrate that the proposed method provides the discrimination ability in classification and annotation, and its performance is better than the other testing methods (sLDA-ann, abc-corr-LDA, SupDocNADE, SAGE and MedSTC) for LabelMe, UIUC, NUS-WIDE and PascalVOC07 images.

An Integration Model of Semantic Annotation Based on Synergetic Neural Network

AbstractCorrect and automatical semantic analysis has always been one of major goals in natural language understanding. However, due to the difficulties in deep semantic analysis, at present, the mainstream studies of semantic analysis are focused on semantic role labeling (SRL) and word sense disambiguation (WSD). Nowadays, these two issues are mostly considered as separate tasks. However, this approach ignores possible dependencies between them. In order to address the issue, an integrative semantic analysis model based on synergetic neural network (SNN) is proposed in this paper, which can easily express useful logic constraints between SRL and WSD. The semantic analysis process can be viewed as the competition process of semantic order parameters. The strongest order parameter will win by competition and desired semantic patterns will be recognized. There are three main innovations in this paper. First, an integrative semantic analysis model is proposed that jointly models word sense disambiguationand...

A novel benchmark model for intelligent annotation of spectral-domain optical coherence tomography scans using the example of cyst annotation

Background and objectivesThe lack of benchmark data in computational ophthalmology contributes to the challenging task of applying disease assessment and evaluate performance of machine learning based methods on retinal spectral domain optical coherence tomography (SD-OCT) scans. Presented here is a general framework for constructing a benchmark dataset for retinal image processing tasks such as cyst, vessel, and subretinal fluid segmentation and as a result, a benchmark dataset for cyst segmentation has been developed. MethodFirst, a dataset captured by different SD-OCT vendors with different numbers of scans and pathology qualities are selected. Then a robust and intelligent method is used to evaluate performance of readers, partitioning the dataset into subsets. Subsets are then assigned to complementary readers for annotation with respect to a novel confidence based annotation protocol. Finally, reader annotations are combined based on their performance to generate final annotations. ResultThe generated benchmark dataset for cyst segmentation comprises 26 SD-OCT scans with differing cyst qualities, collected from 4 different SD-OCT vendors to cover a wide variety of data. The dataset is partitioned into three subsets which are annotated by complementary readers based on a confidence based annotation protocol. Experimental results show annotations of complementary readers are combined efficiently with respect to their performance, generating accurate annotations. ConclusionOur results facilitate the process of generating benchmark datasets. Moreover the generated benchmark data set for cyst segmentation can be used reliably to train and test machine learning based methods.

AggNet: Deep Learning From Crowds for Mitosis Detection in Breast Cancer Histology Images.

The lack of publicly available ground-truth data has been identified as the major challenge for transferring recent developments in deep learning to the biomedical imaging domain. Though crowdsourcing has enabled annotation of large scale databases for real world images, its application for biomedical purposes requires a deeper understanding and hence, more precise definition of the actual annotation task. The fact that expert tasks are being outsourced to non-expert users may lead to noisy annotations introducing disagreement between users. Despite being a valuable resource for learning annotation models from crowdsourcing, conventional machine-learning methods may have difficulties dealing with noisy annotations during training. In this manuscript, we present a new concept for learning from crowds that handle data aggregation directly as part of the learning process of the convolutional neural network (CNN) via additional crowdsourcing layer (AggNet). Besides, we present an experimental study on learning from crowds designed to answer the following questions. 1) Can deep CNN be trained with data collected from crowdsourcing? 2) How to adapt the CNN to train on multiple types of annotation datasets (ground truth and crowd-based)? 3) How does the choice of annotation and aggregation affect the accuracy? Our experimental setup involved Annot8, a self-implemented web-platform based on Crowdflower API realizing image annotation tasks for a publicly available biomedical image database. Our results give valuable insights into the functionality of deep CNN learning from crowd annotations and prove the necessity of data aggregation integration.

PATtyFams: Protein Families for the Microbial Genomes in the PATRIC Database.

The ability to build accurate protein families is a fundamental operation in bioinformatics that influences comparative analyses, genome annotation, and metabolic modeling. For several years we have been maintaining protein families for all microbial genomes in the PATRIC database (Pathosystems Resource Integration Center, patricbrc.org) in order to drive many of the comparative analysis tools that are available through the PATRIC website. However, due to the burgeoning number of genomes, traditional approaches for generating protein families are becoming prohibitive. In this report, we describe a new approach for generating protein families, which we call PATtyFams. This method uses the k-mer-based function assignments available through RAST (Rapid Annotation using Subsystem Technology) to rapidly guide family formation, and then differentiates the function-based groups into families using a Markov Cluster algorithm (MCL). This new approach for generating protein families is rapid, scalable and has properties that are consistent with alignment-based methods.

SeqTools: visual tools for manual analysis of sequence alignments.

BackgroundManual annotation is essential to create high-quality reference alignments and annotation. Annotators need to be able to view sequence alignments in detail. The SeqTools package provides three tools for viewing different types of sequence alignment: Blixem is a many-to-one browser of pairwise alignments, displaying multiple match sequences aligned against a single reference sequence; Dotter provides a graphical dot-plot view of a single pairwise alignment; and Belvu is a multiple sequence alignment viewer, editor, and phylogenetic tool. These tools were originally part of the AceDB genome database system but have been completely rewritten to make them generally available as a standalone package of greatly improved function.FindingsBlixem is used by annotators to give a detailed view of the evidence for particular gene models. Blixem displays the gene model positions and the match sequences aligned against the genomic reference sequence. Annotators use this for many reasons, including to check the quality of an alignment, to find missing/misaligned sequence and to identify splice sites and polyA sites and signals. Dotter is used to give a dot-plot representation of a particular pairwise alignment. This is used to identify sequence that is not represented (or is misrepresented) and to quickly compare annotated gene models with transcriptional and protein evidence that putatively supports them. Belvu is used to analyse conservation patterns in multiple sequence alignments and to perform a combination of manual and automatic processing of the alignment. High-quality reference alignments are essential if they are to be used as a starting point for further automatic alignment generation.ConclusionsWhile there are many different alignment tools available, the SeqTools package provides unique functionality that annotators have found to be essential for analysing sequence alignments as part of the manual annotation process.Electronic supplementary materialThe online version of this article (doi:10.1186/s13104-016-1847-3) contains supplementary material, which is available to authorized users.

Automation of drafting execution by schemes definitions and feature recognition

ABSTRACTDespite of the widespread of annotated 3D virtual models, it is still important to produce engineering drawings for a correct communication among the people of the design and manufacturing chain. Automatic drafting through meta drawing definitions, called Drafting Schemes, is proposed in this paper to save time when drawing becomes a repetitive task. The approach produces drafts with a predetermined content in terms of views arrangement and dimensioning strategy, and guarantees robust results to the variations of the input geometry. In fact, the geometrical entities referenced by the generated dimensions are retrieved through dedicated search algorithms based on recognition of geometric features of the input model.The approach has been implemented and tested on a family of gear motors showing feasibility and good results in terms of drafting time savings.

Maintaining Safety Arguments via Automatic Allocation of Safety Requirements

Abstract: The ‘safety case’ documents the safety argument developers of safety-critical systems employ to convince of their systems’ safety, in compliance with safety standard regulation and advice. Despite the considerable body of knowledge that has evolved, constructing and maintaining a safety case remains a significant challenge. Especially for contemporary systems, due to their scale and complexity, safety cases can grow to require hundreds of pages of documentation. In this paper, we propose a method which aims to address these concerns. In numerous safety standards, such as the aerospace ARP4754-A, the concept of Development Assurance Levels (DALs) is used to control the safety assessment process and influence the safety case. Our method is based on automatically constructing a safety argument from an annotated system architecture model. To perform this construction, we employ previous work towards automatically allocating DALs to such a model and combining it with an appropriate safety argument pattern. The method is enabled through the state-of-the-art model-based dependability tool, HiP-HOPS. The advantage of this approach is that when the design changes, the impact of changes can be automatically reflected in the structure of a re-synthesised safety argument for the system.

Large-scale inference of gene function through phylogenetic annotation of Gene Ontology terms: case study of the apoptosis and autophagy cellular processes.

We previously reported a paradigm for large-scale phylogenomic analysis of gene families that takes advantage of the large corpus of experimentally supported Gene Ontology (GO) annotations. This ‘GO Phylogenetic Annotation’ approach integrates GO annotations from evolutionarily related genes across ∼100 different organisms in the context of a gene family tree, in which curators build an explicit model of the evolution of gene functions. GO Phylogenetic Annotation models the gain and loss of functions in a gene family tree, which is used to infer the functions of uncharacterized (or incompletely characterized) gene products, even for human proteins that are relatively well studied. Here, we report our results from applying this paradigm to two well-characterized cellular processes, apoptosis and autophagy. This revealed several important observations with respect to GO annotations and how they can be used for function inference. Notably, we applied only a small fraction of the experimentally supported GO annotations to infer function in other family members. The majority of other annotations describe indirect effects, phenotypes or results from high throughput experiments. In addition, we show here how feedback from phylogenetic annotation leads to significant improvements in the PANTHER trees, the GO annotations and GO itself. Thus GO phylogenetic annotation both increases the quantity and improves the accuracy of the GO annotations provided to the research community. We expect these phylogenetically based annotations to be of broad use in gene enrichment analysis as well as other applications of GO annotations.Database URL: http://amigo.geneontology.org/amigo

Автоматическая генерация кода по вложенным сетям Петри для систем на основе событий на платформе Telegram

Nested Petri net formalisms is an extension of coloured Petri net formalism that uses Petri Nets as tokens. The formalism allows creating comprehensive models of multi-agent systems, simulating, verifying and analyzing them in a formal and rigorous way. Multi-agent systems are found in many different fields - from safety critical systems to everyday networks of personal computational devices; and, their presence in the real world in increasing with the increasing number of mobile computational devices. While several methods and tools were developed for modelling and analysis of NP-nets models, the synthesis part of multi-agent systems development via NP-nets is still under active development. The widely used method of automatic generation of target system code from designed and verified formal models ensures obtaining correct systems from correct models. In this paper, we demonstrate how Nested Petri net formalism can be applied to model search-and-rescue coordination systems and automatically generate implementation in the form of the executable code for event-driven systems based on the Telegram platform. We augment the NP-nets models with Action Language annotation, which enables us to link transition firings on the model level to Telegram Bot API calls on the implementation level. The suggested approach is illustrated by the example annotated model of a search and rescue coordination system.

Operationalizing mirativity

This study focuses on the conceptual category of mirativity and its constructional construal in English. We propose an operationalization of mirativity with a view to investigating the phenomenon within the usage-based quantitative methodology of multifactorial analysis (Geeraerts, Grondelaers, & Bakema, 1994; Gries, 2003). The proposed operationalization is founded on two usage dimensions, i.e., the degree of performativity of the utterance and the degree of incongruity of the described event. It is argued that mirativity, in its prototypical form, can be operationally defined as a combination of high levels of these two variables. The feasibility of this operationalization in usage-based quantitative research is tested in a case study investigating three surprise-encoding constructions in English: [what + the + np], [what + a + np] and [to + my + np]. The data, amounting to 350 observations of the three constructions, were extracted from dialogic online diaries and submitted to detailed manual annotation and subsequent multivariate statistical modeling. The results reveal a usage continuum ranging from [what + the + np] through [to + my + np] to [what + a + np] relative to the high degrees of performativity and incongruity.

INTEGRATION OF MEDICAL ONTOLOGY CONCEPTS TO ANNOTATE MEDICAL IMAGES

Medical ontology has become an important element to enrich semantic description in medical images. The enrichment process can be done with medical image annotation where label or keyword can be added into the image description. Therefore, in this paper, we have proposed an annotation model that contains two components i.e. image processing and annotation component in order to annotate chest X-ray images with relevant medical ontology concepts. The first component helps transforming the original medical image into six regions of interest for the lung area while the second component prepares medical ontology concepts to annotate the derived regions. By annotating the chest X-ray images with medical ontology concepts, we hope to obtain better and accurate image retrieval.

Two-tier image annotation model based on a multi-label classifier and fuzzy-knowledge representation scheme

Automatic image annotation involves automatically assigning useful keywords to an unlabelled image. The major goal is to bridge the so-called semantic gap between the available image features and the keywords that people might use to annotate images. Although different people will most likely use different words to annotate the same image, most people can use object or scene labels when searching for images.We propose a two-tier annotation model where the first tier corresponds to object-level and the second tier to scene-level annotation. In the first tier, images are annotated with labels of objects present in them, using multi-label classification methods on low-level features extracted from images. Scene-level annotation is performed in the second tier, using the originally developed inference-based algorithms for annotation refinement and for scene recognition. These algorithms use a fuzzy knowledge representation scheme based on Fuzzy Petri Net, KRFPNs, that is defined to enable reasoning with concepts useful for image annotation. To define the elements of the KRFPNs scheme, novel data-driven algorithms for acquisition of fuzzy knowledge are proposed.The proposed image annotation model is evaluated separately on the first and on the second tier using a dataset of outdoor images. The results outperform the published results obtained on the same image collection, both on the object-level and on scene-level annotation. Different subsets of features composed of dominant colours, image moments, and GIST descriptors, as well as different classification methods (RAKEL, ML-kNN and Naïve Bayes), were tested in the first tier. The results of scene level annotation in the second tier are also compared with a common classification method (Naïve Bayes) and have shown superior performance. The proposed model enables the expanding of image annotation with new concepts regardless of their level of abstraction.

GOTA: GO term annotation of biomedical literature.

BackgroundFunctional annotation of genes and gene products is a major challenge in the post-genomic era. Nowadays, gene function curation is largely based on manual assignment of Gene Ontology (GO) annotations to genes by using published literature. The annotation task is extremely time-consuming, therefore there is an increasing interest in automated tools that can assist human experts.ResultsHere we introduce GOTA, a GO term annotator for biomedical literature. The proposed approach makes use only of information that is readily available from public repositories and it is easily expandable to handle novel sources of information. We assess the classification capabilities of GOTA on a large benchmark set of publications. The overall performances are encouraging in comparison to the state of the art in multi-label classification over large taxonomies. Furthermore, the experimental tests provide some interesting insights into the potential improvement of automated annotation tools.ConclusionsGOTA implements a flexible and expandable model for GO annotation of biomedical literature. The current version of the GOTA tool is freely available at http://gota.apice.unibo.it.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-015-0777-8) contains supplementary material, which is available to authorized users.