Use Of Domain-specific Knowledge Research Articles

CS-CO: A Hybrid Self-Supervised Visual Representation Learning Method for H&E-stained Histopathological Images.

Visual representation extraction is a fundamental problem in the field of computational histopathology. Considering the powerful representation capacity of deep learning and the scarcity of annotations, self-supervised learning has emerged as a promising approach to extract effective visual representations from unlabeled histopathological images. Although a few self-supervised learning methods have been specifically proposed for histopathological images, most of them suffer from certain defects that may hurt the versatility or representation capacity. In this work, we propose CS-CO, a hybrid self-supervised visual representation learning method tailored for H&E-stained histopathological images, which integrates advantages of both generative and discriminative approaches. The proposed method consists of two self-supervised learning stages: cross-stain prediction (CS) and contrastive learning (CO). In addition, a novel data augmentation approach named stain vector perturbation is specifically proposed to facilitate contrastive learning. Our CS-CO makes good use of domain-specific knowledge and requires no side information, which means good rationality and versatility. We evaluate and analyze the proposed CS-CO on three H&E-stained histopathological image datasets with downstream tasks of patch-level tissue classification and slide-level cancer prognosis and subtyping. Experimental results demonstrate the effectiveness and robustness of the proposed CS-CO on common computational histopathology tasks. Furthermore, we also conduct ablation studies and prove that cross-staining prediction and contrastive learning in our CS-CO can complement and enhance each other. Our code is made available at https://github.com/easonyang1996/CS-CO.

Using Gene Ontology to Annotate and Prioritize Microarray Data.

The results of high-throughput experiments consist of numerous candidate genes, proteins, or other molecules potentially associated with diseases. A challenge for omics science is the knowledge extraction from the results and the filtering of promising gene or protein candidates. Especially, the hot topic in clinical scenarios consists of highlighting the behavior of few molecules related to some specific disease. In this contest, different computational approaches, also referred Gene prioritization methods, ensure to identify the most related genes to a disease among a larger set of candidate genes. The identification requires the use of domain-specific knowledge that is often encoded into ontologies.

Structural Design via Genetic Optimization

Structural designs (i.e. truss structures) are derived by the use of a three phase genetic optimization approach, where the minimization of volume is the objective of each truss structure considered. A genetic algorithm is employed which controls the three phase optimization technique. The first phase utilizes the conventional functionality of the genetic algorithm from an evolutionary perspective, however designer interaction by the use of constant rules is provided to ensure an effective evolutionary search outcome. The second phase enhances the best design constructed from phase one by the use of domain specific knowledge in the form of design rules. Phase three improves the final design assembled within phase two by the reduction of truss element areas. This refinement process ensures that the design constraints provided are active, indicating an optimal search solution. All phases operate from a global perspective; however the phase two optimization methodology operates from a more radical approach which encompasses the concept of designing from a “blank sheet of paper” point of view. Results are provided upon the conclusion of each truss example considered which includes the outcomes of each phase for comparison purposes.

DropSample: A new training method to enhance deep convolutional neural networks for large-scale unconstrained handwritten Chinese character recognition

Inspired by the theory of Leitner׳s learning box from the field of psychology, we propose DropSample, a new method for training deep convolutional neural networks (DCNNs), and apply it to large-scale online handwritten Chinese character recognition (HCCR). According to the principle of DropSample, each training sample is associated with a quota function that is dynamically adjusted on the basis of the classification confidence given by the DCNN softmax output. After a learning iteration, samples with low confidence will have a higher frequency of being selected as training data; in contrast, well-trained and well-recognized samples with very high confidence will have a lower frequency of being involved in the ongoing training and can be gradually eliminated. As a result, the learning process becomes more efficient as it progresses. Furthermore, we investigate the use of domain-specific knowledge to enhance the performance of DCNN by adding a domain knowledge layer before the traditional CNN. By adopting DropSample together with different types of domain-specific knowledge, the accuracy of HCCR can be improved efficiently. Experiments on the CASIA-OLHDWB 1.0, CASIA-OLHWDB 1.1, and ICDAR 2013 online HCCR competition datasets yield outstanding recognition rates of 97.33%, 97.06%, and 97.51% respectively, all of which are significantly better than the previous best results reported in the literature.

GoD: An R-package based on ontologies for prioritization of genes with respect to diseases

Omics sciences are widely used to analyze diseases at a molecular level. Usually, results of omics experiments are a large list of candidate genes, proteins or other molecules. The interpretation of results and the filtering of candidate genes or proteins selected in an experiment is a challenge in some scenarios. This problem is particularly evident in clinical scenarios in which researchers are interested in the behaviour of few molecules related to some specific disease. The filtering requires the use of domain-specific knowledge that is often encoded into ontologies. To support this interpretation, we implemented GoD (Gene ranking based on Diseases), an algorithm that ranks a given set of genes based on ontology annotations. The algorithm orders genes by the semantic similarity computed with respect to a disease among the annotations of each gene and those describing the selected disease. We tested GoD as proof-of-principle using: Human Phenotype Ontology (HPO), Gene Ontology (GO) and Disease Ontology (DO) using the semantic similarity measures. GoD is publicly available for academic use at https://sites.google.com/site/geneontologyprioritization/.

Semantic Modelling of Interactive 3D Content with Domain-specific Ontologies

The creation of interactive 3D presentations is typically a complex process involving activities related to various aspects of the content such as geometry, structure, space, appearance, logic and behaviour. However, widespread dissemination of interactive 3D content on the web requires flexible and efficient methods of content creation. In this paper, an approach to semantic modelling of 3D content is proposed. The proposed solution enables creation of content components and properties - reflecting different aspects of the content - with domain-specific ontologies and knowledge bases. The use of domain-specific knowledge liberates authors from going into details that are specific to 3D modelling, allows for content representation at different levels of abstraction and permits content creation by domain experts, who are not required to be IT-professionals.

Smart markers for watershed-based cell segmentation.

Automated cell imaging systems facilitate fast and reliable analysis of biological events at the cellular level. In these systems, the first step is usually cell segmentation that greatly affects the success of the subsequent system steps. On the other hand, similar to other image segmentation problems, cell segmentation is an ill-posed problem that typically necessitates the use of domain-specific knowledge to obtain successful segmentations even by human subjects. The approaches that can incorporate this knowledge into their segmentation algorithms have potential to greatly improve segmentation results. In this work, we propose a new approach for the effective segmentation of live cells from phase contrast microscopy. This approach introduces a new set of “smart markers” for a marker-controlled watershed algorithm, for which the identification of its markers is critical. The proposed approach relies on using domain-specific knowledge, in the form of visual characteristics of the cells, to define the markers. We evaluate our approach on a total of 1,954 cells. The experimental results demonstrate that this approach, which uses the proposed definition of smart markers, is quite effective in identifying better markers compared to its counterparts. This will, in turn, be effective in improving the segmentation performance of a marker-controlled watershed algorithm.

The role of domain ontologies in database design

Database design is difficult because it involves a database designer understanding an application and translating the design requirements into a conceptual model. However, the designer may have little or no knowledge about the application or task for which the database is being designed. This research presents a methodology for supporting database design creation and evaluation that makes use of domain-specific knowledge about an application stored in the form of domain ontologies. The methodology is implemented in a prototype system, the Ontology Management and Database Design Environment. Initial testing of the prototype illustrates that the incorporation and use of ontologies is effective in creating entity-relationship models.

Solving SAT by algorithm transform of Wu’s method

Recently algorithms for solving propositional satisfiability problem, or SAT, have aroused great interest, and more attention has been paid to transformation problem solving. The commonly used transformation is representation transform, but since its intermediate computing procedure is a black box from the viewpoint of the original problem, this approach has many limitations. In this paper, a new approach called algorithm transform is proposed and applied to solving SAT by Wu’s method, a general algorithm for solving polynomial equations. By establishing the correspondence between the primitive operation in Wu’s method and clause resolution in SAT, it is shown that Wu’s method, when used for solving SAT, is primarily a restricted clause resolution procedure. While Wu’s method introduces entirely new concepts, e.g. characteristic set of clauses, to resolution procedure, the complexity result of resolution procedure suggests an exponential lower bound to Wu’s method for solving general polynomial equations. Moreover, this algorithm transform can help achieve a more efficient implementation of Wu’s method since it can avoid the complex manipulation of polynomials and can make the best use of domain specific knowledge.

Domain-specific knowledge and memory performance in children

Developmental studies assessing the impact of domain-specific knowledge on memory are discussed. In the first section of the review, different ways through which domain-specific knowledge relates to strategy use in memory tasks are briefly summarized. Empirical evidence indicating nonstrategic effects of the knowledge base are discussed next. In particular, findings based on the expert-novice paradigm are used to compare the knowledge structure and memory performance of experts and novices of different ages, and to explore how individual differences relate to the acquisition and use of domain-specific knowledge. The review shows that domain-specific knowledge permits children to process and remember domain-related information more efficiently, apply strategies more effectively, and integrate novel information more easily than domains for which they have less detailed knowledge. If the knowledge base is particularly rich, it exerts a greater influence than other sources of memory development (i.e., memory capacity, strategies, and metamemory) combined.

USING THE DATA BASE AS A SEMANTIC COMPONENT TO AID IN THE PARSING OF NATURAL LANGUAGE DATA BASE QUERIES

Many of the recent advances in artificial intelligence (Al) have been brought about through the use of domain specific knowledge. In this same spirit, this paper presents an approach to understanding natural language data base queries that employs the use of domain specific knowledge to aid the parsing process. The unique aspect of 1he data base query environment relative to other AI problem domains is that the required body of knowledge already exists in the form of the data base being queried. Thus, there is an important benefit of making use of this knowledge in the form in which it is maintained by the data base management system, since the very difficult problems of gathering and representing this information can be circumvented. The paper describes the problems encountered in the parsing of data base queries that can be solved by the semantic use of the data base, as well as a precise description of how these problems can be solved by existing data base systems. The proposed use of the data base as ...