• Title/Summary/Keyword: combination training

Search Result 406, Processing Time 0.025 seconds

Machine learning-based corporate default risk prediction model verification and policy recommendation: Focusing on improvement through stacking ensemble model (머신러닝 기반 기업부도위험 예측모델 검증 및 정책적 제언: 스태킹 앙상블 모델을 통한 개선을 중심으로)

  • Eom, Haneul;Kim, Jaeseong;Choi, Sangok
    • Journal of Intelligence and Information Systems
    • /
    • v.26 no.2
    • /
    • pp.105-129
    • /
    • 2020
  • This study uses corporate data from 2012 to 2018 when K-IFRS was applied in earnest to predict default risks. The data used in the analysis totaled 10,545 rows, consisting of 160 columns including 38 in the statement of financial position, 26 in the statement of comprehensive income, 11 in the statement of cash flows, and 76 in the index of financial ratios. Unlike most previous prior studies used the default event as the basis for learning about default risk, this study calculated default risk using the market capitalization and stock price volatility of each company based on the Merton model. Through this, it was able to solve the problem of data imbalance due to the scarcity of default events, which had been pointed out as the limitation of the existing methodology, and the problem of reflecting the difference in default risk that exists within ordinary companies. Because learning was conducted only by using corporate information available to unlisted companies, default risks of unlisted companies without stock price information can be appropriately derived. Through this, it can provide stable default risk assessment services to unlisted companies that are difficult to determine proper default risk with traditional credit rating models such as small and medium-sized companies and startups. Although there has been an active study of predicting corporate default risks using machine learning recently, model bias issues exist because most studies are making predictions based on a single model. Stable and reliable valuation methodology is required for the calculation of default risk, given that the entity's default risk information is very widely utilized in the market and the sensitivity to the difference in default risk is high. Also, Strict standards are also required for methods of calculation. The credit rating method stipulated by the Financial Services Commission in the Financial Investment Regulations calls for the preparation of evaluation methods, including verification of the adequacy of evaluation methods, in consideration of past statistical data and experiences on credit ratings and changes in future market conditions. This study allowed the reduction of individual models' bias by utilizing stacking ensemble techniques that synthesize various machine learning models. This allows us to capture complex nonlinear relationships between default risk and various corporate information and maximize the advantages of machine learning-based default risk prediction models that take less time to calculate. To calculate forecasts by sub model to be used as input data for the Stacking Ensemble model, training data were divided into seven pieces, and sub-models were trained in a divided set to produce forecasts. To compare the predictive power of the Stacking Ensemble model, Random Forest, MLP, and CNN models were trained with full training data, then the predictive power of each model was verified on the test set. The analysis showed that the Stacking Ensemble model exceeded the predictive power of the Random Forest model, which had the best performance on a single model. Next, to check for statistically significant differences between the Stacking Ensemble model and the forecasts for each individual model, the Pair between the Stacking Ensemble model and each individual model was constructed. Because the results of the Shapiro-wilk normality test also showed that all Pair did not follow normality, Using the nonparametric method wilcoxon rank sum test, we checked whether the two model forecasts that make up the Pair showed statistically significant differences. The analysis showed that the forecasts of the Staging Ensemble model showed statistically significant differences from those of the MLP model and CNN model. In addition, this study can provide a methodology that allows existing credit rating agencies to apply machine learning-based bankruptcy risk prediction methodologies, given that traditional credit rating models can also be reflected as sub-models to calculate the final default probability. Also, the Stacking Ensemble techniques proposed in this study can help design to meet the requirements of the Financial Investment Business Regulations through the combination of various sub-models. We hope that this research will be used as a resource to increase practical use by overcoming and improving the limitations of existing machine learning-based models.

Transfer Learning using Multiple ConvNet Layers Activation Features with Principal Component Analysis for Image Classification (전이학습 기반 다중 컨볼류션 신경망 레이어의 활성화 특징과 주성분 분석을 이용한 이미지 분류 방법)

  • Byambajav, Batkhuu;Alikhanov, Jumabek;Fang, Yang;Ko, Seunghyun;Jo, Geun Sik
    • Journal of Intelligence and Information Systems
    • /
    • v.24 no.1
    • /
    • pp.205-225
    • /
    • 2018
  • Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.

Progress of Composite Fabrication Technologies with the Use of Machinery

  • Choi, Byung-Keun;Kim, Yun-Hae;Ha, Jin-Cheol;Lee, Jin-Woo;Park, Jun-Mu;Park, Soo-Jeong;Moon, Kyung-Man;Chung, Won-Jee;Kim, Man-Soo
    • International Journal of Ocean System Engineering
    • /
    • v.2 no.3
    • /
    • pp.185-194
    • /
    • 2012
  • A Macroscopic combination of two or more distinct materials is commonly referred to as a "Composite Material", having been designed mechanically and chemically superior in function and characteristic than its individual constituent materials. Composite materials are used not only for aerospace and military, but also heavily used in boat/ship building and general composite industries which we are seeing increasingly more. Regardless of the various applications for composite materials, the industry is still limited and requires better fabrication technology and methodology in order to expand and grow. An example of this is that the majority of fabrication facilities nearby still use an antiquated wet lay-up process where fabrication still requires manual hand labor in a 3D environment impeding productivity of composite product design advancement. As an expert in the advanced composites field, I have developed fabrication skills with the use of machinery based on my past composite experience. In autumn 2011, the Korea government confirmed to fund my project. It is the development of a composite sanding machine. I began development of this semi-robotic prototype beginning in 2009. It has possibilities of replacing or augmenting the exhaustive and difficult jobs performed by human hands, such as sanding, grinding, blasting, and polishing in most often, very awkward conditions, and is also will boost productivity, improve surface quality, cut abrasive costs, eliminate vibration injuries, and protect workers from exposure to dust and airborne contamination. Ease of control and operation of the equipment in or outside of the sanding room is a key benefit to end-users. It will prove to be much more economical than normal robotics and minimize errors that commonly occur in factories. The key components and their technologies are a 360 degree rotational shoulder and a wrist that is controlled under PLC controller and joystick manual mode. Development on both of the key modules is complete and are now operational. The Korean government fund boosted my development and I expect to complete full scale development no later than 3rd quarter 2012. Even with the advantages of composite materials, there is still the need to repair or to maintain composite products with a higher level of technology. I have learned many composite repair skills on composite airframe since many composite fabrication skills including repair, requires training for non aerospace applications. The wind energy market is now requiring much larger blades in order to generate more electrical energy for wind farms. One single blade is commonly 50 meters or longer now. When a wind blade becomes damaged from external forces, on-site repair is required on the columns even under strong wind and freezing temperature conditions. In order to correctly obtain polymerization, the repair must be performed on the damaged area within a very limited time. The use of pre-impregnated glass fabric and heating silicone pad and a hot bonder acting precise heating control are surely required.

Systemic Analysis on Hygiene of Food Catering in Korea (2005-2014) (Systemic analysis 방법을 활용한 국내 학교급식 위생의 주요 영향 인자 분석 연구(2005-2014))

  • Min, Ji-Hyeon;Park, Moon-Kyung;Kim, Hyun-Jung;Lee, Jong-Kyung
    • Journal of Food Hygiene and Safety
    • /
    • v.30 no.1
    • /
    • pp.13-27
    • /
    • 2015
  • A systemic review on the factors affecting food catering hygiene was conducted to provide information for risk management of food catering in Korea. In total 47 keywords relating to food catering and food hygiene were searched for published journals in the DBpia for the last decade (2005-2014). As a result, 1,178 published papers were searched and 142 articles were collected by the expert review. To find the major factors affecting food catering and microbial safety, an analysis based on organization and stakeholder were conducted. School catering (64 papers) was a major target rather than industry (5 pagers) or hospitals (3 papers) in the selected articles. The factors affecting school catering were "system/facility/equipment (15 papers)", "hygiene education (12 papers)", "production/delivery company (6 papers)", food materials (4 papers)" and "any combination of the above factors (9 papers)". The major problems are follow. 1) The problems of "system/facility/equipment" were improper space division/separation, lack of mass cooking utensil, lack of hygiene control equipment, difficulty in temperature and humidity control, and lack of cooperation in the HACCP team (dietitian's position), poor hygienic classroom in the case of class dining (students'), hard workload/intensity of labor, poor condition of cook's safety (cook's) and lack of parents' monitoring activity (parents'). 2) The problem of "hygiene education' were related to formal and perfunctory hygiene education, lack of HACCP education, lack of compliance of hygiene practice (cook's), lack of personal hygiene education and little effect of education (students'). 3) The problems of "production/delivery company" were related to hygiene of delivery truck and temperature control, hygiene of employee in the supplying company and control of non-accredited HACCP company. 4) The area of "food materials" cited were distrust of safety regarding to raw materials, fresh cut produces, and pre-treated food materials. 5) In addition, job stability/the salary can affect the occupational satisfaction and job commitment. And job stress can affect the performance and the hygiene practice. It is necessary for the government to allocate budget for facility and equipment, conduct field survey, improve hygiene training program and inspection, prepare certification system, improve working condition of employees, and introducing hygiene and layout consulting by experts. The results from this study can be used to prepare education programs and develop technology for improving food catering hygiene and providing information.

A Study of the Curriculum Design Modelling Focused on the Combination of National Competency Standards and the Already-Accredited Course in the Department of Social Welfare in the Junior College (과정이수형 자격제도 운영 학과의 NCS 기반 교육과정 설계모형 연구 - 전문대학 사회복지과를 중심으로)

  • Park, Yong Woon;Kim, Kyoung Mee;Yoo, Tae Wan
    • The Journal of the Korea Contents Association
    • /
    • v.16 no.2
    • /
    • pp.652-665
    • /
    • 2016
  • National Competency Standards or NCS is an educational system that emphasizes developing job-related abilities. Therefore it will be an effective solution in training field-oriented work forces if properly applied. However, in the department of social welfare, it is not easy to apply NCS to the curriculum since most academic subjects concerning social welfare focus not on practice but on theory and in addition, most of social welfare departments in junior colleges have an accredited curriculum for the 2nd degree of the social worker qualification. This means it is preposterous if NCS is applied to the curriculum without prior changes in the existing qualification system. So, this paper proposes a draft model to apply NCS to the already-accredited curriculum for the 2nd degree social workers in the junior colleges and details are as follows. Firstly, the competency units will be customized for the existing academic subjects in the curriculum rather than developing new subjects in accordance with NCS competency units. Secondly, some client-related competency units including children, seniors, the disabled are newly developed and then applied to the curriculum, which are crucial for the career development at the junior college level. Thirdly, the competency units are categorized into three types in accordance with the degree of job relevancy - type 1, type 2, type 3. Fourthly, four out of 11 basic job abilities are selected and then developed into academic subjects. Fifthly, all competency units concerning the main job market are regarded as one virtual competency unit and then arranged in the order of type 1s, type 2s and type 3s and then the scope of their study is adjusted to the job abilities required at the main job market.

Effects of Intensive Weight Bearing Treadmill Training and Electrical Stimulation on Skeletal Muscle Properties in Hindlimb Suspended Rats (흰 쥐의 뒷다리 현수 후 집중 체중부하 트레드밀 훈련과 전기자극이 골격근 특성에 미치는 효과)

  • Ahn, Duck-Hyun;Cho, Sang-Hyun;Yi, Chung-Hwi;Kang, Ho-Seok;Kwon, Hyuk-Cheol;Kim, Suhn-Yeop
    • Physical Therapy Korea
    • /
    • v.9 no.1
    • /
    • pp.17-42
    • /
    • 2002
  • The purpose of this study was to investigate the treatment effect of three interventions on the disuse atrophy of rat hindlimb after two weeks suspension. Forty-eight 11~12 weeks old female Sprague-Dawley white rats were divided into four intervention groups: 1) suspension only (S; n=10), 2) intensive weight bearing treadmill (IWBT; n=10), 3) electrical stimulation (ES; n=9), 4) 2)+3) (ES/IWBT; n=9). Another 10 rats received no intervention or hindlimb suspension and served as controls (C). After the interventions, 1) the cross-sectional area (CSA), 2) the ratio of white muscle fiber composition (WMFC), 3) isometric tetanic tension (ITT), and 4) muscle weights (MWs) were measured from the four calf muscle specimens. The results were as follows: 1. In all intervention groups, the CSAs of medial and lateral gastrocnemius (MG LG), soleus (SOL), and flexor digitorum superficialis (FDS) decreased when compared to the control (C) group (p<.05). The CSA increased in FDS and LG for the IWBT group, in SOL for the ES/IWBT group compared to the S only group (p<.05). 2. The ratios of WMFC in MG, LG, SOL, and FDS increased compared to the C group for all interventions (p<.05). The ratios of WMFC decreased in SOL and LG for the IWBT group, in SOL for the ES/IWBT group compared to the S only group, and decreased in SOL for the ES/IWBT group compared to the IWBT group (p<.05). 3. The ITT in the MG, LG, SOL, and FDS decreased compared to the C group for all interventions (p<.05). The ITT increased in MG LG/FDS, SOL, and the whole calf muscles (WCMs) in the IWBT, ES and ES/IWBT groups compared to the S only group (p<.05). 4. The MWs in MG LG/FDS, SOL, WCMs decreased compared to the C group for all interventions (p<.05). The MWs increased in MG LG/FDS and WCMs for the IWBT group, in SOL for the ES group, and in SOL for the ES/IWBT group compared to the S only group (p<.05). 5. In atrophied muscles, the IWBT group showed the best recovery and the ES/IWBT and ES groups followed in decreasing order. The most susceptible muscle to disuse atrophy was the SOL. But conversely, it showed the best recovery in the ES/IWBT group. After two weeks of hindlimb suspension, the calf muscles of rats atrophied and their isometric tension decreased. These changes were best reversed by hindlimb-focused treadmill activity. The next best results were achieved by electrical stimulation combined with the treadmill followed by only electrical stimulation. These findings indicate that full weight bearing treadmill activity alone or in combination with electrical stimulation are effective treatments for non-weight bearing induced muscle atrophy. Further study of the effect of different intensities of electrical stimulation and variations in the duration period of full weight bearing treadmill activity on disuse atrophy is recommended.

  • PDF

Evaluation of Tumor Registry Validity in Samsung Medical Center Radiation Oncology Department (삼성서울병원 방사선종양학과 종양등록 정보의 타당도 평가)

  • Park Won;Huh Seung Jae;Kim Dae Yong;Shin Seong Soo;Ahn Yong Chan;Lim Do Hoon;Kim Seonwoo
    • Radiation Oncology Journal
    • /
    • v.22 no.1
    • /
    • pp.33-39
    • /
    • 2004
  • Purpose : A tumor registry system for the patients treated by radiotherapy at Samsung Medical Center since the opening of a hospital at 1994 was employed. In this study, the tumor registry system was introduced and the validity of the tumor registration was analyzed. Materials and Methods: The tumor registry system was composed of three parts: patient demographic, diagnostic, and treatment Information. All data were input in a screen using a mouse only. Among the 10,000 registered cases in the tumor registry system until Aug, 2002, 199 were randomly selected and their registration data were compared with the patients' medical records. Results : Total input errors were detected on 15 cases (7.5%). There were 8 error items In the part relating to diagnostic Information: tumor site 3, pathology 2, AJCC staging 2 and performance status 1. In the part relating to treatment information there were 9 mistaken items: combination treatment 4, the date of initial treatment 3 and radiation completeness 2. According to the assignment doctor, the error ratio was consequently variable. The doctors who 010 no double-checks showed higher errors than those that 010 (15.6%:3.7%). Conclusion: Our tumor registry had errors within 2% for each Item. Although the overall data qualify was high, further improvement might be achieved through promoting sincerity, continuing training, periodic validity tests and keeping double-checks. Also, some items associated with the hospital Information system will be input automatically In the next step.

Survey of Sedation Practices by Pediatric Dentists (소아치과의사의 진정법 사용에 대한 실태조사)

  • Yang, Yeonmi;Shin, Teojeon;Yoo, Seunghoon;Choi, Seongchul;Kim, Jiyeon;Jeong, Taesung
    • Journal of the korean academy of Pediatric Dentistry
    • /
    • v.41 no.3
    • /
    • pp.257-265
    • /
    • 2014
  • The aim of this study was to establish the appropriate guidelines in the sedation techniques and to organize the continuing education programs for the sedation in future under the direction of Committee on Sedation, Education and Research under the Korean Academy of Pediatric Dentistry(KAPD). The surveys on the sedation technique were performed on 111 organizations which practices the sedation and responded to the survey via online and e-mail by February 2014. The collected survey were analyzed. The purpose of sedation was mainly to manage the children's behavior and its uses were primarily on 3~4 years old children. The most frequent duration of treatment was 1~2 hours to treat both maxillary and mandible. The preferred dosages of sedative drugs were chloral hydrate(CH) 50~70 mg/kg, hydorxyzine(Hx) 1~2 mg/kg, and intramuscular midazolam(Mida IM) 0.1~0.2 mg/kg. The preferred combination of the sedative drugs were CH + Hx + $N_2O/O_2$(67.6%), CH + Hx + Mida submucosal administration (SM) + $N_2O/O_2$(29.7%), and Mida IM + $N_2O/O_2$(23.4%). The administration of additional sedatives was carried out at 48%, mainly using Midazolam. 87.5% of the respondents experienced the adverse effects of the sedation such as vomiting/retching, agitation during recovery, subclinical respiratory depression, staggering, and etc. Among them, only 20% periodically retrain the emergency management protocol. About the discharge criteria for patients after the sedation, the respondents either showed a lack of clear criteria or did not follow the recommended discharge criteria. 86% of the respondents expressed the interests in taking a course on the sedation and they wanted to learn mostly about the sedation-related emergency management, the safe dosage of the sedative drugs, and etc. The use of sedation in pediatric dentistry must be consider a patient's safety as top priority and each dentist must show the evidence of sound practices for the prevention of any possible medical errors. Therefore, KAPD must establish the proper sedation guidelines and it needs to provide the systematic technical training program of sedation-related emergency management for pediatric dentists.

A Study on the Teaching·Learning Management Status and Improvement Plan about 'Creative Engineering Design' Lesson of 'Technology·Home Economics Subject' for High School Teachers (고등학교 기술·가정 교과 '창의 공학 설계' 단원 수업에 대한 교수·학습 운영 실태 분석 및 개선 방안)

  • Kim, SeongIl;Lim, YunJin
    • 대한공업교육학회지
    • /
    • v.41 no.1
    • /
    • pp.128-146
    • /
    • 2016
  • The purpose of this study was to provide a basic research data for increasing the quality of 'creative engineering design' lesson and the teaching learning capability of high school teachers by analyzing the teaching learning management status and improvement plan of 'technology home economics subject' lesson for high school teachers. In order to investigate the teaching learning management status, the survey questionnaires from 63 teachers were collected from high school teachers who teach technology home economics subject currently and analyzed by statistical program SPSS 20. The main results of this study were as follows: First, for the contents of 'creative engineering design' lesson, the highest mean of response was 'creative thinking method'(M=4.22). In the learning activities, the teachers perceived the importance of the 'idea concept' highly. Second, in the management of 'creative engineering design' lesson, the teachers perceived the importance of the secure of tool, material budget, and practice space for the lesson highly. In the teaching capabilities, the teacher perceived the importance of the preparing teaching learning strategy most($$M{\frac{._-}{.}}4.14$$). Third, the teachers preferred to product for making uncomfortable things better in life and the other production outside from the content of textbook. Fourth, in the ratio of practice:theory lesson, they perceived the ratio of 3:7(36.5%), 4:6(25.4%), and 2:8(23.8%) are appropriate. In the assessment, the combination of production, portfolio, and presentation was preferred most. Fifth, there were statistically significant difference in teachers' interest and satisfaction and contents about 'creative engineering design' lesson between groups divided by the existence of practice space, certification held(technology teacher, non technology teacher), etc. Therefore, in order to improve the interest and satisfaction about the 'creative engineering design' lesson, the secure of space for technology practice and material budget were required. In addition, training and seminars program for improving the teaching capability for 'creative engineering design' lesson were required.

The Effect of PNF Technique application Using Thera-Band on the Balance and Gait of Females over 65 years old (세라밴드를 이용한 PNF 기법 적용이 65세 이상 여성노인의 균형과 보행에 미치는 영향)

  • Kang, Dal-Won;Kang, Mi-Kyoung;Kang, Eun-Sil;Go, Yu-Ri;Kim, Da-Woon;Kim, Dae-Yong;Kim, Jung-Eun;Kim, Won-Hwang;Kim, Ja-Yeon;Kim, Hwan;Jung, Dae-In;Kim, Myung-Hoon;Kim, Sang-Yup;Lee, Dong-Jin;Kim, Chan-Kyu;Kim, Hyun-Jin
    • Journal of Korean Physical Therapy Science
    • /
    • v.18 no.1
    • /
    • pp.1-10
    • /
    • 2011
  • Purpose: Study on the effect of the use of Proprioceptive Neuromuscular Facilitation(PNF) method by use of the direction and charge regulation which is the advantage of the Thera-band therapy on the walks and balances in old people and comparison with the result after applying the general PNF technique. Method: The study has been performed on 30 females over 65 years old. The study has been done by dividing the object group in 3 patterns, which are number 1, the comparison group of 10, two ones that are applied the PNF technique using Thera-band and third, the ones that are applied only the PNF technique. For the PNF and the Thera-band using PNF, we have divided the group into Combination of Isotonic technique and the Rhythmical stabilization technique according to the patient's acquaintance pattern and applied them to the patient's body. Evaluation was the balancing ability which was calculated by using the BIODEX Balance system / FRT and for the evaluation of walking ability, we have used the speed of walking for 10M / TUG. Result: In the comparison group of 10, the balancing ability and the walking ability did not change much before and after the experiment, which made it possible to compare the group with the other two easily(p>0.05). For the other two groups, we have recognized the enhancement both in the balancing ability and the walking ability, but they did not know much difference between themselves(p<0.05). Conclusion: Though there were not a big difference in the sense of improvement between the Thera-band using PNF and the PNF technique only, we could infer that these two therapy has enhanced much in the walking and balancing ability for people over 65 and through these result we can foresee that not only using the method shown in this study but also by using many advantages of Thera-band, we could diminish the tiredness of healer, enhance the efficiency of exercise in them and also by forming self training program for older people we could help them build the prevention program from falls.

  • PDF