scholarly journals Knowledge-based Residual Learning

2021 ◽  
Author(s):  
Guanjie Zheng ◽  
Chang Liu ◽  
Hua Wei ◽  
Porter Jenkins ◽  
Chacha Chen ◽  
...  
Keyword(s):  
Domain Knowledge ◽  
Loss Functions ◽  
Superior Performance ◽  
Small Data ◽  
Neural Net ◽  
Knowledge Model ◽  
Residual Learning ◽  
Knowledge Based ◽  
Weak Learner ◽  
Learning Tasks

Small data has been a barrier for many machine learning tasks, especially when applied in scientific domains. Fortunately, we can utilize domain knowledge to make up the lack of data. Hence, in this paper, we propose a hybrid model KRL that treats domain knowledge model as a weak learner and uses another neural net model to boost it. We prove that KRL is guaranteed to improve over pure domain knowledge model and pure neural net model under certain loss functions. Extensive experiments have shown the superior performance of KRL over baselines. In addition, several case studies have explained how the domain knowledge can assist the prediction.

scholarly journals SAR Target Recognition via Meta-Learning and Amortized Variational Inference

Sensors ◽  
2020 ◽  
Vol 20 (20) ◽  
pp. 5966
Author(s):  
Ke Wang ◽  
Gong Zhang
Keyword(s):  
Target Recognition ◽  
Probability Distributions ◽  
Automatic Target Recognition ◽  
Variational Inference ◽  
Training Data ◽  
Superior Performance ◽  
Small Data ◽  
Meta Learning ◽  
Radar Automatic Target Recognition ◽  
Global Parameters

The challenge of small data has emerged in synthetic aperture radar automatic target recognition (SAR-ATR) problems. Most SAR-ATR methods are data-driven and require a lot of training data that are expensive to collect. To address this challenge, we propose a recognition model that incorporates meta-learning and amortized variational inference (AVI). Specifically, the model consists of global parameters and task-specific parameters. The global parameters, trained by meta-learning, construct a common feature extractor shared between all recognition tasks. The task-specific parameters, modeled by probability distributions, can adapt to new tasks with a small amount of training data. To reduce the computation and storage cost, the task-specific parameters are inferred by AVI implemented with set-to-set functions. Extensive experiments were conducted on a real SAR dataset to evaluate the effectiveness of the model. The results of the proposed approach compared with those of the latest SAR-ATR methods show the superior performance of our model, especially on recognition tasks with limited data.

Configuration Tree Solver: A Technology for Automated Design and Configuration

1990 ◽  
Author(s):  
Alexander Kott ◽  
Gerald Agin ◽  
Dave Fawcett
Keyword(s):  
Problem Solving ◽  
General Model ◽  
General Class ◽  
Domain Knowledge ◽  
Automated Design ◽  
Automotive Component ◽  
Business Operation ◽  
Knowledge Based ◽  
Specific Configuration

Abstract Configuration is a process of generating a definitive description of a product or an order that satisfies a set of specified requirements and known constraints. Knowledge-based technology is an enabling factor in automation of configuration tasks found in the business operation. In this paper, we describe a configuration technique that is well suited for configuring “decomposable” artifacts with reasonably well defined structure and constraints. This technique may be classified as a member of a general class of decompositional approaches to configuration. The domain knowledge is structured as a general model of the artifact, an and-or hierarchy of the artifact’s elements, features, and characteristics. The model includes constraints and local specialists which are attached to the elements of the and-or-tree. Given the specific configuration requirements, the problem solving engine searches for a solution, a subtree, that satisfies the requirements and the applicable constraints. We describe an application of this approach that performs configuration and design of an automotive component.

Predicting readers’ domain knowledge based on eye-tracking measures

2018 ◽  
Vol 36 (6) ◽  
pp. 1027-1042 ◽  
Author(s):  
Quan Lu ◽  
Jiyue Zhang ◽  
Jing Chen ◽  
Ji Li
Keyword(s):  
Logistic Regression ◽  
Eye Tracking ◽  
Information Search ◽  
Domain Knowledge ◽  
Fixation Duration ◽  
Baseline Model ◽  
Content Type ◽  
Knowledge Based ◽  
Total Fixation ◽  
Negative Predictor

Purpose This paper aims to examine the effect of domain knowledge on eye-tracking measures and predict readers’ domain knowledge from these measures in a navigational table of contents (N-TOC) system. Design/methodology/approach A controlled experiment of three reading tasks was conducted in an N-TOC system for 24 postgraduates of Wuhan University. Data including fixation duration, fixation count and inter-scanning transitions were collected and calculated. Participants’ domain knowledge was measured by pre-experiment questionnaires. Logistic regression analysis was leveraged to build the prediction model and the model’s performance was evaluated based on baseline model. Findings The results showed that novices spent significantly more time in fixating on text area than experts, because of the difficulty of understanding the information of text area. Total fixation duration on text area (TFD_T) was a significantly negative predictor of domain knowledge. The prediction performance of logistic regression model using eye-tracking measures was better than baseline model, with the accuracy, precision and F(β = 1) scores to be 0.71, 0.86, 0.79. Originality/value Little research has been reported in literature on investigation of domain knowledge effect on eye-tracking measures during reading and prediction of domain knowledge based on eye-tracking measures. Most studies focus on multimedia learning. With respect to the prediction of domain knowledge, only some studies are found in the field of information search. This paper makes a good contribution to the literature on the effect of domain knowledge on eye-tracking measures during N-TOC reading and predicting domain knowledge.

Domain Knowledge-Based Compaction

2013 ◽  
pp. 125-145
Author(s):  
T. Ravindra Babu ◽  
M. Narasimha Murty ◽  
S. V. Subrahmanya
Keyword(s):  
Domain Knowledge ◽  
Knowledge Based

scholarly journals Development of Indonesian language book using glasser model

2017 ◽  
Vol 1 (1) ◽  
pp. 13
Author(s):  
Syamsuyurnita Syamsuyurnita ◽  
Dewi Kesuma Nasution
Keyword(s):  
Teaching And Learning ◽  
Descriptive Analysis ◽  
Learning Objectives ◽  
Learning Activities ◽  
Learning Material ◽  
Teaching Materials ◽  
Study Program ◽  
Descriptive Research ◽  
Knowledge Based ◽  
Learning Tasks

This study aims to describe the process of developing teaching materials by using Glasser model in the Indonesian language course in FKIP UMSU. The sample of the research is 34-second semester A morning students in the Study Program of Language and Literature of Indonesia, Faculty of Teacher Training and Education, University of Muhammadiyah, Sumatera Utara. The questionnaire instrument was used to determine the student's response and activeness to the developed teaching material, the observation sheet used to know the condition of the students in the learning process, and the validation sheet instrument used for the development of teaching materials based on SAP using Glasser model. The result of descriptive research on student's response shows that 100% of students were happy about the teaching materials of Bahasa Indonesia (Teaching Materials, Guided Exercises and Lecture Strategies) and 91.66% of students think that the teaching materials are new to them. After using the teaching materials developed by the researcher and following the teaching and learning activities, students (100%) are interested in following the next lesson, the readability of the language of the learning material is easy to understand (91.66%) and the guidance given by the lecturer is clear (100%). While the self-employed activity is fun for students (91.66%). Students activity in learning activities was shown by their involvement in problem solving, his involvement in carrying out learning tasks, assessing his ability, digging and developing his own knowledge. Based on the validation sheet on the test of learning result 1 obtained information that from the 3 learning objectives formulated in SAP I and SAP II there is 1 learning objectives that have not yet completed. Based on the results of descriptive analysis of the test results of learning 2 it was obtained that the 3 learning objectives formulated in SAP III and SAP IV was finished learning objective.

scholarly journals A Model of Domain Knowledge Content Updating Based on Management Information Interactions

2019 ◽  
Vol 85 ◽  
pp. 69-97
Author(s):  
Jurij Tekutov ◽  
Saulius Gudas ◽  
Vitalijus Denisovas ◽  
Julija Smirnova
Keyword(s):  
Value Chain ◽  
Domain Knowledge ◽  
Chain Model ◽  
Management Information ◽  
Process Measures ◽  
Knowledge Database ◽  
Knowledge Based ◽  
Educational Domain ◽  
Case System ◽  
Knowledge Content

The hierarchical Detailed Value Chain Model and the Elementary Management Cycle model of educational domain knowledge content updating are formally described in this paper, wherein computerized process measures are also proposed. The paper provides a method for updating the knowledge of the analyzed domain, referred to as the “enterprise domain,” based on enterprise modelling in terms of management information interactions. A method was designed, the formal DVCM and EMC descriptions of which are provided in the BPMN notation, allowing to develop a two-level (granular) model for describing the knowledge of educational domain management information interactions. In implementing this model and its algorithms in technological terms, a subsystem of enterprise knowledge has been created in a knowledge-based CASE system (computerized knowledge-based IS engineering), which performs the function of a domain knowledge database.

Sign in / Sign up

Export Citation Format

Share Document