Knowledge-based Representation Research Articles

Dual-Dependency Attention Transformer for Fine-Grained Visual Classification.

Visual transformers (ViTs) are widely used in various visual tasks, such as fine-grained visual classification (FGVC). However, the self-attention mechanism, which is the core module of visual transformers, leads to quadratic computational and memory complexity. The sparse-attention and local-attention approaches currently used by most researchers are not suitable for FGVC tasks. These tasks require dense feature extraction and global dependency modeling. To address this challenge, we propose a dual-dependency attention transformer model. It decouples global token interactions into two paths. The first is a position-dependency attention pathway based on the intersection of two types of grouped attention. The second is a semantic dependency attention pathway based on dynamic central aggregation. This approach enhances the high-quality semantic modeling of discriminative cues while reducing the computational cost to linear computational complexity. In addition, we develop discriminative enhancement strategies. These strategies increase the sensitivity of high-confidence discriminative cue tracking with a knowledge-based representation approach. Experiments on three datasets, NABIRDS, CUB, and DOGS, show that the method is suitable for fine-grained image classification. It finds a balance between computational cost and performance.

Distilling Reliable Knowledge for Instance-Dependent Partial Label Learning

Partial label learning (PLL) refers to the classification task where each training instance is ambiguously annotated with a set of candidate labels. Despite substantial advancements in tackling this challenge, limited attention has been devoted to a more specific and realistic setting, denoted as instance-dependent partial label learning (IDPLL). Within this contex, the assignment of partial labels depends on the distinct features of individual instances, rather than being random. In this paper, we initiate an exploration into a self-distillation framework for this problem, driven by the proven effectiveness and stability of this framework. Nonetheless, a crucial shortfall is identified: the foundational assumption central to IDPLL, involving what we term as partial label knowledge stipulating that candidate labels should exhibit superior confidence compared to non-candidates, is not fully upheld within the distillation process. To address this challenge, we introduce DIRK, a novel distillation approach that leverages a rectification process to DIstill Reliable Knowledge, while concurrently preserves informative fine-grained label confidence. In addition, to harness the rectified confidence to its fullest potential, we propose a knowledge-based representation refinement module, seamlessly integrated into the DIRK framework. This module effectively transmits the essence of similarity knowledge from the label space to the feature space, thereby amplifying representation learning and subsequently engendering marked improvements in model performance. Experiments and analysis on multiple datasets validate the rationality and superiority of our proposed approach.

Knowledge-based representation: Patient engagement in drug development.

Recently, different actors have intensified their efforts to make drug development more participatory. They have produced many frameworks, toolsand dedicated fora, where patients are portrayed as relevant stakeholders to be involved throughout the entire drug development trajectory. To better understand what such participatory efforts entail, in this article, we investigate how patient representation is configured in drug development and what patients can engage as representatives in this field. This is a qualitative study based on the thematic analysis of 40 semistructured interviews with different stakeholders in the field and three patient engagement How-To guides (HTGs) complemented by observations of two sessions of the Patient Engagement Open Forum (PEOF) and a patient expert training of the European Patients' Academy on Therapeutic Innovation (EUPATI). The emerging practices of patient engagement in drug development configure representation as hinging upon three types of knowledge-drug development knowledge, autobiographical knowledgeand community knowledge-and a specific set of skills. We discern a new kind of representation based on these findings, termed 'knowledge-based representation', which appears to more accurately describe how patients are expected to represent others in drug development. Even though knowledge-based representation may be understood as an attempt to downplay the political aspects of representation in favour of its epistemic elements, the political processes involved in patient representation in drug development cannot be ignored. The extent to which reliance on knowledge-based representation will contribute to democratic decision-making is likely to depend on the resources needed to develop the types of knowledge relevant to representation work and on how these types of knowledge are determined. Patient representatives and practitioners in the field of patient engagement (including 13 interviewees, representatives of EUPATIand HTGdevelopers) gave feedback on the interpretation of the findings during a multistakeholder workshop we organised. We also sent an interviewee an extended draft and discussed it during an online meeting. Claudia Egher presented these findings at a PEOF session in June 2023, which further contributed to their validation.

Integration of the Wang & Mendel Algorithm into the Application of Fuzzy Expert Systems to Intelligent Clinical Decision Support Systems

The use of intelligent systems in clinical diagnostics has evolved, integrating statistical learning and knowledge-based representation models. Two recent works propose the identification of risk factors for the diagnosis of obstructive sleep apnea (OSA). The first uses statistical learning to identify indicators associated with different levels of the apnea-hypopnea index (AHI). The second paper combines statistical and symbolic inference approaches to obtain risk indicators (Statistical Risk and Symbolic Risk) for a given AHI level. Based on this, in this paper we propose a new intelligent system that considers different AHI levels and generates risk pairs for each level. A learning-based model generates Statistical Risks based on objective patient data, while a cascade of fuzzy expert systems determines a Symbolic Risk using symptom data from patient interviews. The aggregation of risk pairs at each level involves a fuzzy expert system with automatically generated fuzzy rules using the Wang-Mendel algorithm. This aggregation produces an Apnea Risk indicator for each AHI level, allowing discrimination between OSA and non-OSA cases, along with appropriate recommendations. This approach improves variability, usefulness, and interpretability, increasing the reliability of the system. Initial tests on data from 4400 patients yielded AUC values of 0.74–0.88, demonstrating the potential benefits of the proposed intelligent system architecture.

Knowledge-Based Representation Learning for Nucleus Instance Classification From Histopathological Images.

The classification of nuclei in H&E-stained histopathological images is a fundamental step in the quantitative analysis of digital pathology. Most existing methods employ multi-class classification on the detected nucleus instances, while the annotation scale greatly limits their performance. Moreover, they often downplay the contextual information surrounding nucleus instances that is critical for classification. To explicitly provide contextual information to the classification model, we design a new structured input consisting of a content-rich image patch and a target instance mask. The image patch provides rich contextual information, while the target instance mask indicates the location of the instance to be classified and emphasizes its shape. Benefiting from our structured input format, we propose Structured Triplet for representation learning, a triplet learning framework on unlabelled nucleus instances with customized positive and negative sampling strategies. We pre-train a feature extraction model based on this framework with a large-scale unlabeled dataset, making it possible to train an effective classification model with limited annotated data. We also add two auxiliary branches, namely the attribute learning branch and the conventional self-supervised learning branch, to further improve its performance. As part of this work, we will release a new dataset of H&E-stained pathology images with nucleus instance masks, containing 20,187 patches of size 1024 ×1024 , where each patch comes from a different whole-slide image. The model pre-trained on this dataset with our framework significantly reduces the burden of extensive labeling. We show a substantial improvement in nucleus classification accuracy compared with the state-of-the-art methods.

Cyber-Phishing Website Detection Using Fuzzy Rule Interpolation

This paper introduces a novel detection method for phishing website attacks while avoiding the issues associated with the deficiencies of the knowledge-based representation and the binary decision. The suggested detection method was performed using Fuzzy Rule Interpolation (FRI). The FRI reasoning methods added the benefit of enhancing the robustness of fuzzy systems and effectively reducing the system’s complexity. These benefits help the Intrusion Detection System (IDS) to generate more realistic and comprehensive alerts in case of phishing attacks. The proposed method was applied to an open-source benchmark phishing website dataset. The results show that the proposed detection method obtained a 97.58% detection rate and effectively reduced the false alerts. Moreover, it effectively smooths the boundary between normal and phishing attack traffic because of its fuzzy nature. It has the ability to generate the required security alert in case of deficiencies in the knowledge-based representation. In addition, the results obtained from the proposed detection method were compared with other literature results. The results showed that the accuracy rate of this work is competitive with other methods. In addition, the proposed detection method can generate the required anti-phishing alerts even if one of the anti-phishing sparse rules does not cover some input parameters (observations).

KM[formula omitted]: Visual reasoning via Knowledge Embedding Memory Model with Mutual Modulation

Visual reasoning is a special kind of visual question answering, which is essentially multi-step and compositional, and also requires intensive text-visual interaction. The most important and challenging problem of visual reasoning is to design an effective and robust visual reasoning model. To this end, there are two challenges to overcome. The first is that textual and visual information must be jointly considered to make accurate inferences about reasoning. The second is that existing deep learning-based works are often too specific to a particular task. To address these issues, we propose a knowledge memory embedding model with mutual modulation for visual reasoning. This approach learns not only knowledge-based embeddings derived from key–value memory network to make the full and joint of textual and visual information, but also exploits the prior knowledge to improve the performance with knowledge-based representation learning for applying other general reasoning tasks. Experimental results on four benchmarks show that the proposed approach significantly improves performance compared with other state-of-the-art methods, guarantees the robustness with our model. Most importantly, we apply our model to four reasoning tasks, and experimentally show that our model effectively supports relational reasoning and improves performance in several tasks and datasets.

Integrating Image-Based and Knowledge-Based Representation Learning

A variety of brain areas is involved in language understanding and generation, accounting for the scope of language that can refer to many real-world matters. In this paper, we investigate how regularities among real-world entities impact emergent language representations. Specifically, we consider knowledge bases, which represent entities and their relations as structured triples, and image representations, which are obtained via deep convolutional networks. We combine these sources of information to learn representations of an image-based knowledge representation learning (IKRL) model. An attention mechanism lets more informative images contribute more to the image-based representations. Evaluation results show that the model outperforms all baselines on the tasks of knowledge graph (KG) completion and triple classification. In analyzing the learned models, we found that the structure-based and image-based representations integrate different aspects of the entities and the attention mechanism provides robustness during learning.

Analyzing Global Feature for Duplicate Video Retrieval using CNN and PCA

At present, Duplicate video retrieval has attained the interest of researchers because of the vast amount of online videos. This video retrieval has extensive applications like online video monitoring, copyright protection, and automatic video tagging. Some local features constitute primary building blocks in this video retrieval algorithms; with this, most researchers use local information for feature representation. Moreover, this local knowledge-based representation eliminates more prominent information regarding global distribution. However, the discriminative power of local descriptors is diminished by feature quantifiers. The ultimate goal is to use universal features to categorize similar keyframes into the same class that is essential to enhance video retrieval performance. Here, CNN features acquire global geometric distribution of video, from which discrete features are considered for computation. Discretization is performed with principal component analysis. These kinds of features maintain geometry transformation with reduced noise. Next, an integration strategy known as k-NN is used to merge these features with global VR features for enhancing recognition accuracy. Experimentation has been carried out with available datasets to show that the anticipated model outperforms existing approaches in VR applications

Knowledge-Based Coding of Objects for Multisource Surveillance Video Data

Global object redundancy (GOR), as opposed to local spatial/temporal redundancies in a single video clip, is a new form of redundancy common in multisource surveillance video data (MSVD). GOR is induced by the repetition of foreground objects across multiple cameras, and becomes influential as the number of objects increases. Eliminating GOR considerably improves MSVD coding efficiency. In an effort to accomplish this, this study first proposes a knowledge-based representation of objects based on careful analysis of GOR composition. The representation contains a constant part and a variational part: the former is used to represent the common knowledge shared by an object across multiple cameras, while the latter is used to represent local variations on the object's surfaces. Based on the proposed representation, a knowledge-based coding (KBC) method is then proposed in which each foreground object is encoded with a hybrid prediction scheme, where the constant part of the object is generated via global prediction from a model library and the variational part is predicted via local reference frames with pose-based, short-term prediction. Experimental results showed that the KBC method saves more than 39% bits on average for encoding foreground objects in high-resolution video clips (compared to 16% for the entire videos). Applying the proposed coding method to surveillance videos in large spatial and temporal scale allows storage savings at the PB level.

HARMONIZATION BETWEEN PERSONAL AND SHARED MEMORIES

This paper discusses harmonization issues between the management of resources stored locally by users in personal repositories and the management of resources shared by a community which they belong to. The solution consists in maintaining the same kind of indexing. We discuss some requirements that the motivation implies and we present some choices that have been done for engineering the system that allows to manage different types of resources and that allows to describe them with ontological elements. This paper focuses on a knowledge-based representation of resources that supports the service that the community can bring to its members. We develop our approach of semantic indexing. Some constraints due to the nature of a distributed index are also presented. Peer to peer based communities rely on a decentralized system. The paper also presents the system engineering of a prototype that gives access to the resources. It explains how the architecture is structured and the role of its layers. The Web design of the front-end guarantees an anywhere and anytime access to the resources. The system supports an easy scalability thanks to its underlying architecture based on web services.

An operational model based on knowledge representation for querying the image content with concepts and relations

In order to overcome the semantic gap (i.e. the gap between low-level extracted features and semantic description in state-of-the-art content-based image retrieval systems, a class of frameworks proposed within the framework of the European Fermi project, consisted of modeling the semantic content of images following a sharp process of human-assisted indexing. These approaches, based on expressive knowledge-based representation models provide satisfactory results in terms of retrieval quality but are not easily usable on large collections of images because of the necessary human intervention required for indexing. We propose in this paper to integrate the content-based and semantic-based solutions through a model featuring semantic and relational characterizations of the multimedia (image) content for automatic symbolic indexing and retrieval. Its instantiation as an image retrieval framework relies on a representation formalism handling high-level image descriptions and allowing to query with conceptual descriptors. Our approach complements content-based solutions through the mapping of low-level extracted features to semantic concepts and the manipulation of graph-based symbolic index and query structures; and extends the semantic-based solutions by considering automatically-extracted semantic and relational information. At the experimental level, we evaluate the retrieval performance of our system on queries coupling both semantic and relational characterizations through recall and precision indicators on a test collection of 2,500 color photographs and the TRECVID keyframe corpus.

Empirical Validation of the RCDC and RCDE Semantic Complexity Metrics for Object-Oriented Software

The Relative Class Domain Complexity (RCDC) and Relative Class Definition Entropy (RCDE) semantic metrics have been proposed for use as complexity metrics for object-oriented software. These semantic metrics are calculated on a knowledge-based representation of software, following a knowledge-based program understanding examination of the software. The metrics have great potential because they can be applied during the software design phase whereas most complexity metrics cannot be applied until after development is complete. In this paper, we present the results of a study to empirically validate the RCDC and RCDE metrics. We show that the metrics compare favorably with the findings of human experts and also that they correlate well with the results of conventional complexity metrics.

Software maintenance project delays prediction using Bayesian Networks

Managing software maintenance is rarely a precise task due to uncertainties concerned with resources and services descriptions. Even when a well-established maintenance process is followed, the risk of delaying tasks remains if the new services are not precisely described or when resources change during process execution. Also, the delay of a task at an early process stage may represent a different delay at the end of the process, depending on complexity or services reliability requirements. This paper presents a knowledge-based representation (Bayesian Networks) for maintenance project delays based on specialists experience and a corresponding tool to help in managing software maintenance projects.

Predicting Class II MHC-Peptide binding: a kernel based approach using similarity scores

BackgroundModelling the interaction between potentially antigenic peptides and Major Histocompatibility Complex (MHC) molecules is a key step in identifying potential T-cell epitopes. For Class II MHC alleles, the binding groove is open at both ends, causing ambiguity in the positional alignment between the groove and peptide, as well as creating uncertainty as to what parts of the peptide interact with the MHC. Moreover, the antigenic peptides have variable lengths, making naive modelling methods difficult to apply. This paper introduces a kernel method that can handle variable length peptides effectively by quantifying similarities between peptide sequences and integrating these into the kernel.ResultsThe kernel approach presented here shows increased prediction accuracy with a significantly higher number of true positives and negatives on multiple MHC class II alleles, when testing data sets from MHCPEP [1], MCHBN [2], and MHCBench [3]. Evaluation by cross validation, when segregating binders and non-binders, produced an average of 0.824 AROC for the MHCBench data sets (up from 0.756), and an average of 0.96 AROC for multiple alleles of the MHCPEP database.ConclusionThe method improves performance over existing state-of-the-art methods of MHC class II peptide binding predictions by using a custom, knowledge-based representation of peptides. Similarity scores, in contrast to a fixed-length, pocket-specific representation of amino acids, provide a flexible and powerful way of modelling MHC binding, and can easily be applied to other dynamic sequence problems.

A Granular Description of ECG Signals

In this paper, we develop a general framework of a granular representation of ECG signals. The crux of the approach lies in the development and ongoing processing realized in the setting of information granules-fuzzy sets. They serve as basic conceptual and semantically meaningful entities using which we describe signals and build their models (such as various predictive schemes or classifiers). A comprehensive two-phase scheme of the design of the information granules is proposed and described. At the first phase, we discuss the temporal granulation through a series of temporal windows (granular windows) and an aggregation of the values of signal by means of fuzzy sets. To address this issue, offered is a detailed method of building a fuzzy set based on numeric data and a certain optimization criterion that strikes a balance between the highest experimental relevance of the fuzzy set supported by numeric data and its substantial specificity. At the next phase of the granular design, a collection of information granules is further summarized with the use of fuzzy clustering (Fuzzy C-Means). The resulting prototypes (centroids) formed by this grouping process serve as elements of the granular vocabulary. We discuss ways of using these vocabularies in the knowledge-based representation, modeling, and classification of ECG beats.

An artificial intelligence approach for generating assembly sequences in CAD/CAM

This paper presents an innovative artificial intelligence approach for generating assembly sequences on a consortium of database emulating expert systems. Procedures include shape and feature recognition using a model-based CAD (computer-aided design) analyser, data structure and data modeling, knowledge-based representation, and inference processing through a set of heuristics and rules. The main tool here is an object-oriented concept as a means for managing geometrical data, topological data and abstraction. Abstractions added to data semantics help to build a knowledge base to meet assembly requirements. As a result, an AI paradigm supports knowledge processing for the search of good solutions and good decisions in the assembly process. An efficient implementation of the integrated system has been achieved and successfully demonstrated by results based on parts with varying complexities.

Evaluating Knowledge-Based Statistical Reasoning

Comparisons of reasoning by novices and experts generally suggest that as knowledge increases, general strategies are replaced by the application of systems of knowledge-based concept representation. To assess how training in statistics is associated with the development of “expert” thinking, the relation between general and statistical critical thinking ability was assessed in 50 undergraduates. For students who scored above the median on a final examination of statistical knowledge, formal training was associated with an increase in statistical reasoning and the disappearance of the prior relation between general and statistical reasoning. Students with low knowledge showed neither improved statistical reasoning nor a relation between general and statistical critical thinking either before or after training.

Knowledge-Based Medical Image Analysis and Representation for Integrating Content Definition with the Radiological Report

Technology breakthroughs in high-speed, high-capacity, and high performance desk-top computers and workstations make the possibility of integrating multimedia medical data to better support clinical decision making, computer-aided education, and research not only attractive, but feasible. To systematically evaluate results from increasingly automated image segmentation it is necessary to correlate them with the expert judgments of radiologists and other clinical specialists interpreting the images. These are contained in increasingly computerized radiological reports and other related clinical records. But to make automated comparison feasible it is necessary to first ensure compatibility of the knowledge content of images with the descriptions contained in these records. Enough common vocabulary, language, and knowledge representation components must be represented on the computer, followed by automated extraction of image-content descriptions from the text, which can then be matched to the results of automated image segmentation. A knowledge-based approach to image segmentation is essential to obtain the structured image descriptions needed for matching against the expert's descriptions. We have developed a new approach to medical image analysis which helps generate such descriptions: a knowledge-based object-centered hierarchical planning method for automatically composing the image analysis processes. The problem-solving steps of specialists are represented at the knowledge level in terms of goals, tasks, and domain objects and concepts separately from the implementation level for specific representations of different image types, and generic analysis methods. This system can serve as a major functional component in incrementally building and updating a structured and integrated hybrid information system of patient data.(ABSTRACT TRUNCATED AT 250 WORDS)

A KNOWLEDGE-BASED FRAMEWORK OF OBJECT-ORIENTED SOFTWARE DEVELOPMENT ENVIRONMENTS

Software development can be seen as a formalization process which, from informal data (problem), leads to a fully formal representation (solution) by successive enrichments and transformations. This paper presents a representation formalism (KLSDE: Knowledge-based representation Language for Software Development Environments) which supports the iterative and evolutionary features of this process. From syntactic and semantic rules described in the domain model and the software process model, the representation is gradually bound by stronger and stronger constraints, according to the degree of formalization. Based on object concepts, KLSDE offers both expressive power as well as simplicity and readability. The power, which stems from knowledge representation techniques, is necessary to build various models such as the software process development, the domain or the solution. Readability and simplicity are required so that the language addresses communication needs. For readability, the language is constructed on current metaphors and concepts. It includes a graphic representation and many structuring mechanisms. Finally, KLSDE is at the center of a development environment. It is parameterized by the user methodology and fulfills the needs for the whole life-cycle.