Domain Expert Knowledge Research Articles

Clinical Report Guided Retinal Microaneurysm Detection With Multi-Sieving Deep Learning.

Timely detection and treatment of microaneurysms is a critical step to prevent the development of vision-threatening eye diseases such as diabetic retinopathy. However, detecting microaneurysms in fundus images is a highly challenging task due to the low image contrast, misleading cues of other red lesions, and the large variation of imaging conditions. Existing methods tend to fail in face of the large intra-class variation and small inter-class variations for microaneurysm detection in fundus images. Recently, hybrid text/image mining computer-aided diagnosis systems have emerged to offer a promise of bridging the semantic gap between images and diagnostic information. In this paper, we focus on developing an interleaved deep mining technique to cope intelligently with the unbalanced microaneurysm detection problem. Specifically, we present a clinical report guided multi-sieving convolutional neural network, which leverages a small amount of supervised information in clinical reports to identify the potential microaneurysm regions via the image-to-text mapping in the feature space. These potential microaneurysm regions are then interleaved with fundus image information for multi-sieving deep mining in a highly unbalanced classification problem. Critically, the clinical reports are employed to bridge the semantic gap between low-level image features and high-level diagnostic information. We build an efficient microaneurysm detection framework based on the hybrid text/image interleaving and validate its performance on challenging clinical data sets acquired from diabetic retinopathy patients. Extensive evaluations are carried out in terms of fundus detection and classification. Experimental results show that our framework achieves 99.7% precision and 87.8% recall, comparing favorably with the state-of-the-art algorithms. Integration of expert domain knowledge and image information demonstrates the feasibility of reducing the difficulty of training classifiers under extremely unbalanced data distributions.

Scale Up Event Extraction Learning via Automatic Training Data Generation

The task of event extraction has long been investigated in a supervised learning paradigm, which is bound by the number and the quality of the training instances. Existing training data must be manually generated through a combination of expert domain knowledge and extensive human involvement. However, due to drastic efforts required in annotating text, the resultant datasets are usually small, which severally affects the quality of the learned model, making it hard to generalize. Our work develops an automatic approach for generating training data for event extraction. Our approach allows us to scale up event extraction training instances from thousands to hundreds of thousands, and it does this at a much lower cost than a manual approach. We achieve this by employing distant supervision to automatically create event annotations from unlabelled text using existing structured knowledge bases or tables.We then develop a neural network model with post inference to transfer the knowledge extracted from structured knowledge bases to automatically annotate typed events with corresponding arguments in text.We evaluate our approach by using the knowledge extracted from Freebase to label texts from Wikipedia articles. Experimental results show that our approach can generate a large number of highquality training instances. We show that this large volume of training data not only leads to a better event extractor, but also allows us to detect multiple typed events.

Interactive medical word sense disambiguation through informed learning.

Medical word sense disambiguation (WSD) is challenging and often requires significant training with data labeled by domain experts. This work aims to develop an interactive learning algorithm that makes efficient use of expert's domain knowledge in building high-quality medical WSD models with minimal human effort. We developed an interactive learning algorithm with expert labeling instances and features. An expert can provide supervision in 3 ways: labeling instances, specifying indicative words of a sense, and highlighting supporting evidence in a labeled instance. The algorithm learns from these labels and iteratively selects the most informative instances to ask for future labels. Our evaluation used 3 WSD corpora: 198 ambiguous terms from Medical Subject Headings (MSH) as MEDLINE indexing terms, 74 ambiguous abbreviations in clinical notes from the University of Minnesota (UMN), and 24 ambiguous abbreviations in clinical notes from Vanderbilt University Hospital (VUH). For each ambiguous term and each learning algorithm, a learning curve that plots the accuracy on the test set against the number of labeled instances was generated. The area under the learning curve was used as the primary evaluation metric. Our interactive learning algorithm significantly outperformed active learning, the previous fastest learning algorithm for medical WSD. Compared to active learning, it achieved 90% accuracy for the MSH corpus with 42% less labeling effort, 35% less labeling effort for the UMN corpus, and 16% less labeling effort for the VUH corpus. High-quality WSD models can be efficiently trained with minimal supervision by inviting experts to label informative instances and provide domain knowledge through labeling/highlighting contextual features.

Similarity Measurement of Metadata of Geospatial Data: An Artificial Neural Network Approach

To help users discover the most relevant spatial datasets in the ever-growing global spatial data infrastructures (SDIs), a number of similarity measures of geospatial data based on metadata have been proposed. Researchers have assessed the similarity of geospatial data according to one or more characteristics of the geospatial data. They created different similarity algorithms for each of the selected characteristics and then combined these elementary similarities to the overall similarity of the geospatial data. The existing combination methods are mainly linear and may not be the most accurate. This paper reports our experiences in attempting to learn the optimal non-linear similarity integration functions, from the knowledge of experts, using an artificial neural network. First, a multiple-layer feed forward neural network (MLFFN) was created. Then, the intrinsic characteristics were used to represent the metadata of geospatial data and the similarity algorithms for each of the intrinsic characteristics were built. The training and evaluation data of MLFFN were derived from the knowledge of domain experts. Finally, the MLFFN was trained, evaluated, and compared with traditional linear combination methods, which was mainly a weighted sum. The results show that our method outperformed the existing methods in terms of precision. Moreover, we found that the combination of elementary similarities of experts to the overall similarity of geospatial data was not linear.

Using echo state networks for classification: A case study in Parkinson's disease diagnosis.

Despite having notable advantages over established machine learning methods for time series analysis, reservoir computing methods, such as echo state networks (ESNs), have yet to be widely used for practical data mining applications. In this paper, we address this deficit with a case study that demonstrates how ESNs can be trained to predict disease labels when stimulated with movement data. Since there has been relatively little prior research into using ESNs for classification, we also consider a number of different approaches for realising input-output mappings. Our results show that ESNs can carry out effective classification and are competitive with existing approaches that have significantly longer training times, in addition to performing similarly with models employing conventional feature extraction strategies that require expert domain knowledge. This suggests that ESNs may prove beneficial in situations where predictive models must be trained rapidly and without the benefit of domain knowledge, for example on high-dimensional data produced by wearable medical technologies. This application area is emphasized with a case study of Parkinson's disease patients who have been recorded by wearable sensors while performing basic movement tasks.

Playing style recognition through an adaptive video game

Playing style recognition is crucially important for style-based adaptation of digital games. Unlike traditional ways for measuring of styles by means of self-reports, automatic style estimation incorporated into a video game appears to be a more efficient and ecologically valid method. The article presents a model for in-game recognition of four playing styles (Competitor, Dreamer, Logician, and Strategist) based on the Kolb's experiential learning theory. The model applies multiple linear regression over task performance metrics as explanatory variables and coefficients found first by a heuristic approach relaying on experience and observation knowledge of domain experts and, next, estimated by the least squares method. Experiments with the model implemented within an affectively adaptive video game demonstrated the benefits of emotion-based dynamic difficulty adjustment over playing outcomes and proved its validity as an accurate instrument for automatic estimation of both the four playing styles and the learning styles of Honey and Mumford.

Intelligent wear mode identification system for marine diesel engines based on multi-level belief rule base methodology

Wear faults are among the chief causes of main-engine damage, significantly influencing the secure and economical operation of ships. It is difficult for engineers to utilize multi-source information to identify wear modes, so an intelligent wear mode identification model needs to be developed to assist engineers in diagnosing wear faults in diesel engines. For this purpose, a multi-level belief rule base (BBRB) system is proposed in this paper. The BBRB system consists of two-level belief rule bases, and the 2D and 3D characteristics of wear particles are used as antecedent attributes on each level. Quantitative and qualitative wear information with uncertainties can be processed simultaneously by the BBRB system. In order to enhance the efficiency of the BBRB, the silhouette value is adopted to determine referential points and the fuzzy c-means clustering algorithm is used to transform input wear information into belief degrees. In addition, the initial parameters of the BBRB system are constructed on the basis of expert-domain knowledge and then optimized by the genetic algorithm to ensure the robustness of the system. To verify the validity of the BBRB system, experimental data acquired from real-world diesel engines are analyzed. Five-fold cross-validation is conducted on the experimental data and the BBRB is compared with the other four models in the cross-validation. In addition, a verification dataset containing different wear particles is used to highlight the effectiveness of the BBRB system in wear mode identification. The verification results demonstrate that the proposed BBRB is effective and efficient for wear mode identification with better performance and stability than competing systems.

Exploiting load testing and profiling for Performance Antipattern Detection

Context: The performance assessment of complex software systems is not a trivial task since it depends on the design, code, and execution environment. All these factors may affect the system quality and generate negative consequences, such as delays and system failures. The identification of bad practices leading to performance flaws is of key relevance to avoid expensive rework in redesign, reimplementation, and redeployment.Objective: The goal of this manuscript is to provide a systematic process, based on load testing and profiling data, to identify performance issues with runtime data. These performance issues represent an important source of knowledge as they are used to trigger the software refactoring process. Software characteristics and performance measurements are matched with well-known performance antipatterns to document common performance issues and their solutions.Method: We execute load testing based on the characteristics of collected operational profile, thus to produce representative workloads. Performance data from the system under test is collected using a profiler tool to create profiler snapshots and get performance hotspot reports. From such data, performance issues are identified and matched with the specification of antipatterns. Software refactorings are then applied to solve these performance antipatterns.Results: The approach has been applied to a real-world industrial case study and to a representative laboratory study. Experimental results demonstrate the effectiveness of our tool-supported approach that is able to automatically detect two performance antipatterns by exploiting the knowledge of domain experts. In addition, the software refactoring process achieves a significant performance gain at the operational stage in both case studies.Conclusion: Performance antipatterns can be used to effectively support the identification of performance issues from load testing and profiling data. The detection process triggers an antipattern-based software refactoring that in our two case studies results in a substantial performance improvement.

An Introduction to Deep Learning for the Physical Layer

We present and discuss several novel applications of deep learning for the physical layer. By interpreting a communications system as an autoencoder, we develop a fundamental new way to think about communications system design as an end-to-end reconstruction task that seeks to jointly optimize transmitter and receiver components in a single process. We show how this idea can be extended to networks of multiple transmitters and receivers and present the concept of radio transformer networks as a means to incorporate expert domain knowledge in the machine learning model. Lastly, we demonstrate the application of convolutional neural networks on raw IQ samples for modulation classification which achieves competitive accuracy with respect to traditional schemes relying on expert features. This paper is concluded with a discussion of open challenges and areas for future investigation.

워드임베딩과 그래프 기반 준지도학습을 통한 한국어 어휘 감성 점수 산출

Sentiment analysis plays an important role in both public and private sectors to understand consumers’ responses to products or voters’ reactions to policies. One of the most key success factors of sentiment analysis is to build an appropriate sentiment word dictionary. Many current existing approaches either heavily rely on the knowledge of domain experts or word co-occurrence statistics, the first of which causes low efficiency and high expenditure while the second of which suffers from incomplete data. In order to resolve these shortcomings, we propose a new domain-specific Korean word sentiment score evaluation method based on word embedding and graph based semi-supervised learning. First, words are embedded in a lower dimensional space by Word2Vec technique. Then, the word relation graph is constructed based on the similarity between words in the embedding space. Then, we assign sentiments to approximately 1% words utilizing some indicators like centrality measure. The sentiment scores of the other unlabeled words are automatically assigned by label propagation with semisupervised learning. To verify our proposed method, we collect 1.98 million review comments from three movie review websites. Experimental results show that the proposed method achieves about 93% accuracy of polarity classification.

Natural Model based Design in Context: an Effective Method for Environmental Problems

Analyzing complex problem domains is not easy. Simulation tools support decision makers to find the best policies. Model-based system development is an approach where a model of the application domain is the central driving force when designing simulation tools. State-of-the-art techniques however still require both expert knowledge of the application domain and the implementation techniques as provided by ICT (such as multilevel agent technology). Domain experts, however, usually do not master ICT sufficiently. Modeling is more insightful for the domain expert when its goal is to formalize the language being used in that domain as a semi-natural language. At the meta level, this language describes the main concepts of the type of application domain. The model then is a concretization of this meta model. The main focus of this article is (1) to propose a natural-language-based approach to modeling application domains, (2) to show how these models can be transformed systematically into computational models, and (3) to propose the tool TiC (Tool in Context) that supports the domain expert when developing a model and generating a simulation tool. Our research methodology is based on Design Science. We verify our approach by describing the various transformation steps in detail, and by demonstrating the way of working via a sample session applying a real problem of Laf Forest Reserve deforestation in North Cameroon.

Hyperdimensional Computing for Blind and One-Shot Classification of EEG Error-Related Potentials

The mathematical properties of high-dimensional (HD) spaces show remarkable agreement with behaviors controlled by the brain. Computing with HD vectors, referred to as “hypervectors,” is a brain-inspired alternative to computing with numbers. HD computing is characterized by generality, scalability, robustness, and fast learning, making it a prime candidate for utilization in application domains such as brain–computer interfaces. We describe the use of HD computing to classify electroencephalography (EEG) error-related potentials for noninvasive brain–computer interfaces. Our algorithm naturally encodes neural activity recorded from 64 EEG electrodes to a single temporal–spatial hypervector without requiring any electrode selection process. This hypervector represents the event of interest, can be analyzed to identify the most discriminative electrodes, and is used for recognition of the subject’s intentions. Using the full set of training trials, HD computing achieves on average 5% higher single-trial classification accuracy compared to a conventional machine learning method on this task (74.5% vs. 69.5%) and offers further advantages: (1) Our algorithm learns fast: using only 34% of training trials it achieves an average accuracy of 70.5%, surpassing the conventional method. (2) Conventional method requires prior domain expert knowledge, or a separate process, to carefully select a subset of electrodes for a subsequent preprocessor and classifier, whereas our algorithm blindly uses all 64 electrodes, tolerates noises in data, and the resulting hypervector is intrinsically clustered into HD space; in addition, most preprocessing of the electrode signal can be eliminated while maintaining an average accuracy of 71.7%.

Transforming BPMN 2.0 Business Process Model into SBVR Business Vocabulary and Rules

As the success of information system projects and their development also relies on the knowledge of domain experts, the modelled processes should be presented to them for validation purposes before the implementation stage. Domain experts do not have the knowledge of business process modelling notations. Therefore, the validation may not be correct. However, they better understand structured natural language, such as SBVR that defines the meaning of business concepts and make them unambiguously understandable by human experts and also by software systems. The solution presented in this paper allows transforming BPMN 2.0 business process models into SBVR business vocabularies and business rules. This solution is implemented as a plug-in and is available in the MagicDraw CASE tool. An experimental evaluation of the proposed solution with three domain areas proved that SBVR business vocabulary and business rules could be fully obtained from BPMN 2.0 business process model when certain requirements for BPMN 2.0 business process model are met. The advantages of the solution are an automatic transformation for various BPMN modelling situations, tracing links between two models in one modelling environment and avoidance of necessity to use linguistic techniques. DOI: http://dx.doi.org/10.5755/j01.itc.46.2.18520

Combining human and machine intelligence to derive agents’ behavioral rules for groundwater irrigation

Abstract For agent-based modeling, the major challenges in deriving agents’ behavioral rules arise from agents’ bounded rationality and data scarcity. This study proposes a “gray box” approach to address the challenge by incorporating expert domain knowledge (i.e., human intelligence) with machine learning techniques (i.e., machine intelligence). Specifically, we propose using directed information graph (DIG), boosted regression trees (BRT), and domain knowledge to infer causal factors and identify behavioral rules from data. A case study is conducted to investigate farmers' pumping behavior in the Midwest, U.S.A. Results show that four factors identified by the DIG algorithm- corn price, underlying groundwater level, monthly mean temperature and precipitation- have main causal influences on agents’ decisions on monthly groundwater irrigation depth. The agent-based model is then developed based on the behavioral rules represented by three DIGs and modeled by BRTs, and coupled with a physically-based groundwater model to investigate the impacts of agents’ pumping behavior on the underlying groundwater system in the context of coupled human and environmental systems.

Semantic Enrichment for Building Information Modeling: Procedure for Compiling Inference Rules and Operators for Complex Geometry

Semantic enrichment of building models adds meaningful domain-specific or application-specific information to a digital building model. It is applicable to solving interoperability problems and to compilation of models from point cloud data. The SeeBIM (Semantic Enrichment Engine for BIM) prototype software encapsulates domain expert knowledge in computer readable rules for inference of object types, identity and aggregation of systems. However, it is limited to axis-aligned bounding box geometry and the adequacy of its rule-sets cannot be guaranteed. This paper solves these drawbacks by (1) devising a new procedure for compiling inference rule sets that are known a priori to be adequate for complete and thorough classification of model objects, and (2) enhancing the operators to compute complex geometry and enable precise topological rule processing. The procedure for compiling adequate rule sets is illustrated using a synthetic concrete highway bridge model. A real-world highway bridge model, with 333 components of 13 different types and compiled from a laser scanned point cloud, is used to validate the approach and test the enhanced SeeBIM system. All of the elements are classified correctly, demonstrating the efficacy of the approach to semantic enrichment.

Interactive robogami: An end-to-end system for design of robots with ground locomotion

This paper aims to democratize the design and fabrication of robots, enabling people of all skill levels to make robots without needing expert domain knowledge. Existing work in computational design and rapid fabrication has explored this question of customization for physical objects but so far has not been able to conquer the complexity of robot designs. We have developed Interactive Robogami, a tool for composition-based design of ground robots that can be fabricated as flat sheets and then folded into 3D structures. This rapid prototyping process enables users to create lightweight, affordable, and materially versatile robots with short turnaround time. Using Interactive Robogami, designers can compose new robot designs from a database of print-and-fold parts. The designs are tested for the users’ functional specifications via simulation and fabricated on user satisfaction. We present six robots designed and fabricated using a 3D printing based approach, as well as a larger robot cut from sheet metal. We have also conducted a user study that demonstrates that our tool is intuitive for novice designers and expressive enough to create a wide variety of ground robot designs.

Technology Analysis of Global Smart Light Emitting Diode (LED) Development Using Patent Data

Technological developments related to smart light emitting diode (LED) systems have progressed rapidly in recent years. In this paper, patent documents related to smart LED technology are collected and analyzed to understand the technology development of smart LED systems. Most previous studies of the technology were dependent on the knowledge and experience of domain experts, using techniques such as Delphi surveys or technology road-mapping. These approaches may be subjective and lack robustness, because the results can vary according to the selected expert groups. We therefore propose a new technology analysis methodology based on statistical modeling to obtain objective and relatively stable results. The proposed method consists of visualization based on Bayesian networks and a linear count model to analyze patent documents related to smart LED technology. Combining these results, a global hierarchical technology structure is created that can enhance the sustainability in smart LED system technology. In order to show how this methodology could be applied to real-world problems, we carry out a case study on the technology analysis of smart LED systems.

A convolutional neural network based feature learning and fault diagnosis method for the condition monitoring of gearbox

Feature extraction plays a vital role in intelligent fault diagnosis of mechanical system. Nevertheless, traditional feature extraction methods suffer from three problems, which are (1) the requirements of domain expertise and prior knowledge, (2) the sensitive to the changes of mechanical system and (3) the limitations of mining new features. It is attractive and meaningful to investigate an automatic feature extraction method, which can adaptively learn features from raw data and discover new fault-sensitive features. Deep learning has been widely used in image analysis and speech recognition with great success. The key advantage of this method lies into the ability of mining representative information and sensitive features from raw data. However, the application of deep learning in feature leaning for mechanical diagnosis is still few, and limited studies have been carried out to compare the effectiveness of feature leaning with various data types. This paper will focus on developing a convolutional neural network (CNN) to learn features directly from frequency data of vibration signals and testing the different performance of feature learning from raw data, frequency spectrum and combined time-frequency data. Manual features from time domain, frequency domain and wavelet domain as well as three common intelligent methods are used as comparisons. The effectiveness of the proposed method is validated through PHM 2009 gearbox challenge data and a planetary gearbox test rig. The results demonstrate that the proposed method is able to learn features adaptively from frequency data and achieve higher diagnosis accuracy than other comparative methods.

ActiveMotif: Interactive motif discovery with human feedback.

Motif detection, which is to discover short patterns involved in many important biological processes, has been recently raised as an important task in bioinformatics. The traditional algorithms to find a sequence motif have been developed using machine learning only without involving the experience and domain knowledge of human experts effectively. In this paper, we propose an interactive motif discovery system by introducing a new learning algorithm, by generalizing a well-known statistical motif model, whose inference can be shepherded by human feedback.

Effect of Condition Monitoring on Risk Mitigation for Steam Turbines in the Forest Products Industry

Industry has been implementing condition monitoring (CM) for turbines to minimize losses and to improve productivity. Deficient conditions can be identified before losses occur by monitoring the equipment parameters. For any loss scenario, the effectiveness of monitoring depends on the stage of the loss scenario when the deficient condition is detected. A scenario-based semi-empirical methodology was developed to assess various types of condition monitoring techniques, by considering their effect on the risk associated with mechanical breakdown of steam turbines in the forest products (FPs) industry. A list of typical turbine loss scenarios was first generated by reviewing loss data and leveraging expert domain knowledge. Subsequently, condition monitoring techniques that can mitigate the risk associated with each loss scenario were identified. For each loss scenario, an event tree analysis (ETA) was used to quantitatively assess the variations in the outcomes due to condition monitoring, and resultant changes in the risk associated with turbine mechanical breakdown. An application was developed following the methodology to evaluate the effect of condition monitoring on turbine risk mitigation.