The sensory stimulation of a cosmetic product has been deemed to be an ancillary aspect until a decade ago. That point of view has drastically changed on different levels in just a decade. Nowadays cosmetic formulators should unavoidably meet the needs of consumers who want sensory satisfaction, although they do not have much time for new product development. The selection of new products from candidate products largely depend on the panel of human sensory experts. As new product development cycle time decreases, the formulators wanted to find systematic tools that are required to filter candidate products into a short list. Traditional statistical analysis on most physical property tests for the products including tribology tests and rheology tests, do not give any sound foundation for filtering candidate products. In this paper, we suggest a deep learning-based analysis method to identify hand cream products by raw electric signals from tribological sliding test. We compare the result of the deep learning-based method using raw data as input with the results of several machine learning-based analysis methods using manually extracted features as input. Among them, ResNet that is a deep learning model proved to be the best method to identify hand cream used in the test. According to our search in the scientific reported papers, this is the first attempt for predicting test cosmetic product with only raw time-series friction data without any manual feature extraction. Automatic product identification capability without manually extracted features can be used to narrow down the list of the newly developed candidate products.

Recently, transfer learning techniques with a base convolutional neural network (CNN) model have widely gained acceptance in early detection and classification of crop diseases to increase agricultural productivity with reducing disease spread. The transfer learning techniques based classifiers generally achieve over 90% of classification accuracy for crop diseases using dataset of crop leaf images (e.g., PlantVillage dataset), but they have ability to classify only the pre-trained diseases. This paper provides with an evaluation scheme on selecting an effective base CNN model for crop disease transfer learning with regard to the accuracy of trained target crops as well as of untrained target crops. First, we present transfer learning models called CDC (crop disease classification) architecture including widely used base (pre-trained) CNN models. We evaluate each performance of seven base CNN models for four untrained crops. The results of performance evaluation show that the DenseNet201 is one of the best base CNN models.

This research examines deep learning based image recognition models for beef sirloin classification. The sirloin of beef can be classified as the upper sirloin, the lower sirloin, and the ribeye, whereas during the distribution process they are often simply unified into the sirloin region. In this work, for detailed classification of beef sirloin regions we develop a model that can learn image information in a reasonable computation time using the MobileNet algorithm. In addition, to increase the accuracy of the model we introduce data augmentation methods as well, which amplifies the image data collected during the distribution process. This data augmentation enables to consider a larger size of training data set by which the accuracy of the model can be significantly improved. The data generated during the data proliferation process was tested using the MobileNet algorithm, where the test data set was obtained from the distribution processes in the real-world practice. Through the computational experiences we confirm that the accuracy of the suggested model is up to 83%. We expect that the classification model of this study can contribute to providing a more accurate and detailed information exchange between suppliers and consumers during the distribution process of beef sirloin.

목적 : 본 연구는 델파이 기법을 사용하여 지역사회 임상에서 치매환자의 중증도를 파악하기 위한 포괄적인 인지기능을 평가할 수 있는 항목을 개발하고자 한다. 연구방법 : 지역사회 치매환자의 중증도 파악을 위한 예비문항 구성과 치매 전문가 패널 21명을 대상으로 3회에 걸쳐 델파이 조사를 실시하였다. 예비문항 구성은 문헌고찰을 통한 항목을 수집하였다. 1차 조사에서는 수집된 항목에 대한 폐쇄형과 개방형 문항으로 전문가 패널들의 의견을 수집하였다. 2차, 3차 조사에서는 항목에 대한 내용타당도(Content Validity Ratio; CVR)를 조사하고, 최종 항목은 CVR, 수렴 도, 합의도 기준을 모두 만족하는 항목을 선정하여 도출하였다. 결과 : 문헌고찰을 통해 지역사회 임상에서 치매환자의 임상적인 기능을 평가할 수 있는 예비 평가항목의 하위영역 5개, 문항 59개를 도출하였다. 1차 델파이 결과에서 3개 항목 삭제, 16개 항목이 추가되어 68 개 항목이 선정되었다. 2차 델파이 결과에서는 내용타당도 비율 .42 미만인 항목 3개가 있었으나 문헌 고찰과 연구자 간의 합의를 통해 3차 델파이 조사에 포함시켰다. 3차 델파이 결과, 수렴도 .50 이상과 합의도 .75 미만인 항목 9개를 삭제하여 총 59개 항목을 선정하였다. 최종 델파이 결과, 각 항목에 대한 내용타당도는 .89, 안정도는 .12, 수렴도는 .43, 합의도는 .81로 높게 분석되었다. 결론 : 본 연구를 통해 작업치료사가 지역사회에 거주하고 있는 치매환자의 중증도를 파악하기 위하여 임상적인 치매 기능평가를 할 수 있는 항목들을 개발하고 평가항목에 대한 내용타당도를 검증하였다. 지역 사회 작업치료 종사자가 치매환자의 치매 기능을 임상적으로 파악하고 중증도에 따라 적절한 중재를 제공하기 위한 기초자료로 활용되기를 기대한다.

PURPOSES : This study aims to develop and evaluate computer vision-based algorithms that classify the road roughness index (IRI) of road specimens with known IRIs. The presented study develops and compares classifier-based and deep learning-based models that can effectively determine pavement roughness grades. METHODS : A set road specimen was developed for various IRIs by generating road profiles with matching standard deviations. In addition, five distinct features from road images, including mean, peak-to-peak, standard variation, and mean absolute deviation, were extracted to develop a classifier-based model. From parametric studies, a support vector machine (SVM) was selected. To further demonstrate that the model is more applicable to real-world problems, with a non-integer road grade, a deep-learning model was developed. The algorithm was proposed by modifying the MNIST database, and the model input parameters were determined to achieve higher precision. RESULTS : The results of the proposed algorithms indicated the potential of using computer vision-based models for classifying road surface roughness. When SVM was adopted, near 100% precision was achieved for the training data, and 98% for the test data. Although the model indicated accurate results, the model was classified based on integer IRIs, which is less practical. Alternatively, a deep-learning model, which can be applied to a non-integer road grade, indicated an accuracy of over 85%. CONCLUSIONS : In this study, both the classifier-based, and deep-learning-based models indicated high precision for estimating road surface roughness grades. However, because the proposed algorithm has only been verified against the road model with fixed integers, optimization and verification of the proposed algorithm need to be performed for a real road condition.

In horse management, the alarm system with sensors in the foaling period enables the breeder can appropriately prepare the time of the parturition. It is important to prevent losses by unpredictable parturition because there are several high risks such as dystocia and the death of foals and mares during foaling. However, unlike analysis in the alarm system that detects specific motions has been widely performed, analysis of classification following specific behavior patterns or number needs to be more organized. Thus, the objective of this study is to classify signs of the specific behaviors of the mares for the prediction of pre-foaling behaviors. Five Thoroughbred mares (9-20 yrs) were randomly selected for observation of the prefoaling behaviors. The behaviors were monitored for 90 min that was divided into three different periods as 1) from -90 to -60 min, 2) from -60 to -30 min, 3) from -30 min to the time for the discharge of the amniotic fluid, respectively. The behaviors were divided into two different categories as state and frequent behaviors and each specific behavioral pattern for classification was individually described. In the state behaviors, the number of mares in the standing of the foaling group (3.17 ± 0.18b) at period 3 was significantly higher than the control group (1.67 ± 0.46a). In contrast, the number of the mares in the eating of the foaling group (1.17 ± 0.34b) at period 3 was significantly lower than the control group (3.33 ± 0.46a). In the frequent behaviors, the weaving of the foaling group was significantly higher than the control group, and looking at the belly of the foaling group was significantly lower than the control group. In period 2, defecation, weaving, and lowering the head of the foaling group were significantly higher than the control group, respectively. In period 3, sitting down and standing up, pawing, weaving, and lowering the head in the foaling group were also significantly higher than the control group. In conclusion, the behavior is significantly different in foaling periods, and the prediction of foaling may be feasible by the detection of the pre-foaling behaviors in the mares.

자기공명영상은 고해상도의 연부조직에 대한 영상정보를 제공하며, 뇌종양 등 연부조직 진단에 활용된다. 본 연구는 합성곱신경망 인공지능을 통해 뇌종양 자기공명영상 분류성능을 확인해 보고자 한다. 4개 종류로 구분된 3264 장의 MRI 데이터 세트(data set)를 이용하였으며, 인공지능 학습을 위해 훈련용 데이터와 시험용 데이터를 9 : 1, 훈련용 데이터의 10%를 검증용 데이터로 구분하였다. 합성곱신경망은 기본 CNN과 VGG16으로 구성하였으며, 학습 평가는 정확도와 손실율로 확인하였으며, 생성된 모델을 통해 분류성능 정확도를 확인하였다. 실험 결과 과적합은 없었으며, 분류성능은 기본 CNN과 VGG16 각각 67%와 80%의 분류성능을 보였다. 도출된 뇌종양 자기공명영상 분류 결과를 통해 자기공명영상과 인공지능 접목에 관한 기초 자료로 사용될 수 있을 것이라 사료된다.

MRI는 연부조직에 대한 고해상도의 영상을 제공하며 진단적 가치가 매우 높은 영상 검사이며, 디지털 데이터를 이용하여 딥러닝 기술을 통해 컴퓨터 보조 진단 역할을 수행할 수 있다. 본 연구는 딥러닝 기반 YOLOv3를 이용하여 뇌종양 분류 성능을 확인해 보고자 한다. 253장의 오픈 MRI 영상을 이용하여 딥러닝 학습을 진행하고 학습 평가지표는 평균손실(average loss)와 region 82와 region 94를 사용하였으며, 뇌종양 분류 모델 검증을 위해 학습에 사용되지 않은 영상을 이용하여 검출 성능을 평가하였다. 평균손실은 2248 epochs 시 0.1107, region 82와 region 94의 24079 반복학습 시 average IoU, class, .5R, .75R은 각각 0.89와 0.81, 1.00과 1.00, 1.00과 1.00, 1.00과 1.00의 결과값을 도출하였다. 뇌종양 분류 모델 검증 결과 정상 뇌와 뇌종양 각각 95.00%, 75.36%의 정확도로 분류할 수 있었다. 본 연구 결과를 통해 MRI 영상을 활용한 딥러닝 연구 및 임상에 기초자료로 사용될 것이라 사료된다.

최근 소프트웨어 유통망(ESD)를 통해 유통되는 게임이 늘어남에 따라 사용자들이 원하는 게임을 찾는데 어려움을 겪고 있다. 이에 따라 사용자들이 원하는 게임을 찾기 쉽게 태그를 생성하는 모델의 필요성이 대두 되고 있다. 본 논문에서는 태그를 생성하는 모델을 BERT를 통해 설계하였다. 태그 100개 중 가장 적합한 태그를 4개 추출하기 위해 입력된 문장에 대해 각 태그별로 이진분류를 수행하고 이진분류 당시의 Softmax 값이 가장 컸던 태그 4개를 선택했다. 또한, 모델의 정확도를 위해서 약 33억 개의 다국어 단어로 학습한 pre-trained Multilingual BERT 모델과 약 5천만 개의 한국어 단어로 학습한 KoBERT 모델을 가져와 한국어 데이터로 학습(finetuning) 시켜 사용하였다. 실험에서 BERT 모델은 KoBERT 모델보다 F- 점수에서 9.19 % 더 나은 성능을 보입니다. 이는 언어 학습 데이터 세트의 크기가 특정 언어인 한국어 특성보다 더 중요하다는 것을 나타낸다.

Metal three-dimensional (3D) printing is an important emerging processing method in powder metallurgy. There are many successful applications of additive manufacturing. However, processing parameters such as laser power and scan speed must be manually optimized despite the development of artificial intelligence. Automatic calibration using information in an additive manufacturing database is desirable. In this study, 15 commercial pure titanium samples are processed under different conditions, and the 3D pore structures are characterized by X-ray tomography. These samples are easily classified into three categories, unmelted, well melted, or overmelted, depending on the laser energy density. Using more than 10,000 projected images for each category, convolutional neural networks are applied, and almost perfect classification of these samples is obtained. This result demonstrates that machine learning methods based on X-ray tomography can be helpful to automatically identify more suitable processing parameters.