• International journal of advanced smart convergence
• /
• 제5권3호
• /
• pp.1-7
• /
• 2016

In existing Convolutional Neural Network (CNNs) for object recognition task, there are only few efforts known to reduce the noises from the images. Both convolution and pooling layers perform the features extraction without considering the noises of the input image, treating all pixels equally important. In computer vision field, there has been a study to weight a pixel importance. Seam carving resizes an image by sacrificing the least important pixels, leaving only the most important ones. We propose a new way to combine seam carving approach with current existing CNN model for object recognition task. We attempt to remove the noises or the "unimportant" pixels in the image before doing convolution and pooling, in order to get better feature representatives. Our model shows promising result with CIFAR-10 dataset.

• 아동학회지
• /
• 제23권3호
• /
• pp.123-137
• /
• 2002

This study investigated children's cognitions of physically impossible events by characteristics of the events. The 67 subjects were 3-, 5-, and 7-year old children from one day care center and one elementary school in Seoul. Children's responses to questions regarding the events were taped and transcribed. Cognition of the events was measured by recognition of impossibility, possibility of replication, and reasons for the possibility of replication of the events. Data were analyzed by ANOVA(repeated measures), and Pearson's correlation. Children's recognition of the impossibility of the events varied by type of task : Recognition of gravity tasks was different from irreversibility tasks, but there were no differences between recognition of irreversibility and gravity and between irreversibility and object permanence. The possibility of replication correlated significantly with reasons given by the children.

• Journal of the Korean Physical Society
• /
• 제73권12호
• /
• pp.1908-1917
• /
• 2018

The binding affinity between the T-cell receptors (TCRs) and antigenic peptides mainly determines immunological recognition. It is not a trivial task that T cells identify the digital sequences of peptide amino acids by simply relying on the integrated binding affinity between TCRs and antigenic peptides. To address this problem, we examine whether the affinity-based discrimination of peptide sequences is learnable and generalizable by artificial neural networks (ANNs) that process the digital experimental amino acid sequence information of receptors and peptides. A pair of TCR and peptide sequences correspond to the input for ANNs, while the success or failure of the immunological recognition correspond to the output. The output is obtained by both theoretical model and experimental data. In either case, we confirmed that ANNs could learn the immunological recognition. We also found that a homogenized encoding of amino acid sequence was more effective for the supervised learning task.

• 동의생리병리학회지
• /
• 제20권2호
• /
• pp.471-476
• /
• 2006

Chongmyung-tang(CMT, 聰明湯), oriental herbal medicine which consists of Polygaglae Radix(遠志), Acori Graminei Rhizoma(石菖蒲) and Hoelen(白茯神) has effect on amnesia, dementia. In order to evaluate effect of CMT on memory and learning in mice, CMT extract was used for studies. This paper describes the effects of CMT extract on memory and learning processes by using the passive and active avoidance performance tests, novel object recognition task and water maze task. The CMT extract ameliorated the memory retrieval deficit induced by ethanol in the passive avoidance responses but did not affect ambulatory activity of normal mice. These results suggest that CMT has an ameliorating effect on memory retrieval impairment. CMT extract decreased spontaneous motor activity(SMA) in the latter sessions of memory registration in active avoidance responses. These results suggest that CMT has partly transquilizing or antianxiety effects. In novel object recognition task to measure visual recognition memory, CMT-administered mice enhanced in long term memory for 1-3 days. In water maze task to measure spatial learning, which requires the activation of NMDA receptors in the hippocampus, spatial learning in CMT-administered mice was faster than in wild-type mice. These results suggest that CMT enhances memory and activates NMDA receptors.

• 한국심리학회지 : 문화 및 사회문제
• /
• 제11권4호
• /
• pp.83-95
• /
• 2005

본 연구는 상황인식의 첫 번째 단계인 지각, 특히 운전상황과 관련한 대부분의 정보를 획득하는 공간지각 능력과 상황인식의 다음 단계인 이해와 예측 사이의 관계를 검토하였다. 실제 도로상황을 편집한 동영상으로 구성된 실험 재료를 이용해 재인과 예측 능력을 측정하였으며, 이 두 가지 요소들을 통합하는 과제로 운전 시뮬레이터를 조작하면서 간단한 숫자 배열 규칙에 따라 결과를 계산(예측)하는 과제를 사용하였다. 본 연구의 결과를 요약하면 다음과 같다. 첫째, 운전-관련 상황에서 오퍼레이터의 공간지각 능력이 우수할수록 실제 도로상황 재인과제 수행의 민감도가 높았다. 둘째, 공간지각 능력이 좋을수록 실제 도로상황 예측과제에서의 예측률이 높았다. 마지막으로 공간지각능력이 우수할수록 이해와 예측을 통합적으로 요구되는 숫자-계산 과제에서의 수행이 우수하였다. 본 연구 결과, 운전자 상황인식 능력의 측정방법으로 공간지각능력 검사의 활용을 제안할 수 있으며, 비교적 간단한 절차인 계산검사를 통해 상황인식의 이해와 예측을 통합적으로 살펴볼 수 있음을 시사한다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 한국자동제어학술회의논문집(국내학술편); 포항공과대학교, 포항; 24-26 Oct. 1996
• /
• pp.516-519
• /
• 1996

Sub-assembly in ship manufacturing is a sequence of filet joint welding of stiffeners on metal panels and the process is different depending on companies. In this paper, we introduce a new intelligent robotic system of the sub-assembly process in Samsung Heavy Industry, where one shift of 22m * 9m workspace includes one to ten panels and each panel includes up to 10 stiffeners. The inherent problems such as several hundreds of different panels, unstructured task environment and the large scale do not allow a fixed automation, but needs highly intelligent versatile automation. The robotic system is composed of four 14DOF macro-mini robots and a task recognition system. Application of this system has verified the task specification such as low temperature environment(-10.deg. C) and productivity is satisfied successfully.

• 인지과학
• /
• 제23권2호
• /
• pp.225-248
• /
• 2012

시각단기재인 과정에 대한 유사성 합산 모형은 단기기억에 저장된 기억 표상들과 재인검사 자극간 유사성의 합산치가 재인의사결정을 좌우한다고 제안한다[1]. 본 연구는 수직 공간주파수를 달리하는 두 Gabor 자극에 대한 공간주파수 재인 과제에서 과제와 무관한 기억자극의 위치 정보가 재인 반응에 미치는 영향을 조사하였다. 실험의 매 시행에서 참가자는 시야 좌우에 제시된 서로 다른 두 Gabor 자극의 공간주파수를 기억하고 약 1초 뒤에 제시된 재인검사 자극과의 동일성 여부를 보고하였다. 재인검사 자극은 기억 항목과 동일하거나 기억된 항목을 기준으로 공간주파수 차원에서 1 또는 4 최소식별단위 만큼 표적과 유사한 미끼자극이었다. 또한 재인검사항목이 제시되는 위치는 재인이 요구된 기억항목의 위치를 기준으로 동일하거나 혹은 반대편 시야 위치에 제시되었으며, 참가자는 검사자극의 위치변화를 무시하고 공간주파수 차원의 재인 반응만을 요구받았다. 실험 결과 미끼자극에 대한 오인반응(false recognition)은 기억항목과 미끼자극 간 유사성 합산치가 커짐에 따라 증가했다. 아울러 과제와는 무관하게 검사항목의 위치 변화가 있었던 조건의 경우 없었던 조건에 비해 이러한 오인 반응의 빈도가 감소한 것으로 나타났다. 과제 무관련 위치 변화에 의한 미끼자극에 대한 오인 반응의 감소는 재인과제의 요구와는 관계없이 무관련 세부특징이 과제 관련 세부특징에 대한 재인 반응에 영향을 미칠 수 있음을 보여주며, 시각단기기억에 저장되는 정보가 세부특징들의 독립적 표상이 아닌 세부특징 간 공고한 결합에 의해 표상될 가능성을 시사한다.

• 방송공학회논문지
• /
• 제24권1호
• /
• pp.58-66
• /
• 2019

대화 문장 내 고유명사와 같은 개체명에 대한 인식 연구는 효율적 대화 정보 예측을 위한 가장 기본적이며 중요한 연구 분야이다. 목적 지향 대화 시스템에서 가장 주요한 부분은 대화 내 객체가 어떤 속성을 가지고 있느냐 하는 것을 인지하는 것이다. 개체명 인식모델은 대화 문장에 대하여 전처리, 단어 임베딩, 예측 단계를 통해 개체명 인식을 진행한다. 본 연구는 효율적인 대화 정보 예측을 위해 전처리 단계에서 사용자 정의 사전을 이용하고 단어 임베딩 단계에서 최적의 파라미터를 발견하는 것을 목표로 한다. 그리고 설계한 개체명 인식 모델을 실험하기 위해 생활 화학제품 분야를 선택하고 관련 도메인 내 목적 지향 대화 시스템에서 적용 할 수 있는 개체명 인식 모델을 구축하였다.

• 재활간호학회지
• /
• 제3권2호
• /
• pp.181-195
• /
• 2000

The purpose of this study is to find out the effects of personal characteristics, social support, depression, and metamemory on the older adults' memory performances. The subjects of the study consisted of 95 older adults over the age of 60 who are living in Busan. Some data were collected by means of the interview method, using questionnaires for metamemory (MIA questionnaire by Dixon, et al., 1988), and depression(GDS by Yesavage and Sheikl, 1986), social resoueces(Duke university, 1978), family support(Hyun-Sook Kang, 1985). The other data were collected by the testing method on the memory performance such as the immeadiate word recall task, the delayed word recall task, the word recognition task(Elderly Verbal Learning Test by Kyung Mi Choi, 1998), and the face recognition task(Face Recognition Task tool developed by Hye-Sook Min). The results of this study were as follows; 1. The level of metamemory is 3.4 points in the 5 point scale, the grades of the task and the achievement are relatively high and the grades of the change, the control, and the anxiety are relatively low in the sub-concepts. 2. Metamemory have significant relation with age(r=-.44), educational attainment(r=.46), depression(r=-.58), family support(r=.20), social resources(r=.20). 3. The significant variables to predict older adults' metamemory are educational attainment(22%), sex(11%), age (8.3%), depression(4.7%), and illness state(3%). 4. The strong variables to predict memory performances are metamemory, age, depression, social resources, educational attainment, illness state, and limitation of daily living activity related to illness. In conclusion, the enhancement strategies of metamemory and the social support and the prevention or reduction of depression are necessary to increase older adults' memory performances. Ultimately in this respect nurses' roles are very important in developing and performing some intervention programs for old adults' memory improvement, which have significant meanings in the field of nursing science.

• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 2003년도 ISIS 2003
• /
• pp.480-483
• /
• 2003

We propose a new framework that aims at multi-purpose image recognition, a difficult task for the conventional rule-based systems. This framework is farmed based on the idea of computer-based learning algorithm. In this research, we introduce the new functions of an additional learning and a knowledge reconstruction on the Fuzzy Inference Neural Network (FINN) (1) to enable the system to accommodate new objects and enhance the accuracy as necessary. We examine the capability of the proposed framework using two examples. The first one is the capital letter recognition task from UCI machine learning repository to estimate the effectiveness of the framework itself, Even though the whole training data was not given in advance, the proposed framework operated with a small loss of accuracy by introducing functions of the additional learning and the knowledge reconstruction. The other is the scenery image recognition. We confirmed that the proposed framework could recognize images with high accuracy and accommodate new object recursively.