Deep learning-based computer vision anomaly detection algorithms are widely utilized in various fields. Especially in the manufacturing industry, the difficulty in collecting abnormal data compared to normal data, and the challenge of defining all potential abnormalities in advance, have led to an increasing demand for unsupervised learning methods that rely on normal data. In this study, we conducted a comparative analysis of deep learning-based unsupervised learning algorithms that define and detect abnormalities that can occur when transparent contact lenses are immersed in liquid solution. We validated and applied the unsupervised learning algorithms used in this study to the existing anomaly detection benchmark dataset, MvTecAD. The existing anomaly detection benchmark dataset primarily consists of solid objects, whereas in our study, we compared unsupervised learning-based algorithms in experiments judging the shape and presence of lenses submerged in liquid. Among the algorithms analyzed, EfficientAD showed an AUROC and F1-score of 0.97 in image-level tests. However, the F1-score decreased to 0.18 in pixel-level tests, making it challenging to determine the locations where abnormalities occurred. Despite this, EfficientAD demonstrated excellent performance in image-level tests classifying normal and abnormal instances, suggesting that with the collection and training of large-scale data in real industrial settings, it is expected to exhibit even better performance.

도로의 포장 상태의 노후화나 관리미흡으로 인하여 시민의 사유 재산 중 주요한 요소인 자동차 등의 손상이나 자동차 사고 로 이어질 수 있어 큰 사회적 비용이 발생할 뿐 아니라, 시민들의 불편과 불만을 초래할 수 있다. 최근 도로 포장의 경우 포트홀 발생 건수와 그에 따른 민원 및 소송 건수가 증가해 행정력 및 예산이 낭비되고 있으며, 서울시의 경우 포장도로 노후화 추이가 증가함에 따라 유 지 관리 비용 또한 증가하고 있다. SOC 시설물 안전성 강화에 대한 사회적 요구는 지속적으로 증가하고 있어 한정된 예산의 효율적 활용을 위한 첨단 유지관리기술 도입이 시급하다.

The high-level radioactive waste repository must ensure its performance for a long period of time enough to sufficiently reduce the potential risk of the waste, and for this purpose, multibarrier systems consisting of engineered and natural barrier systems are applied. If waste nuclides leak, the dominating mechanisms facilitating their movement toward human habitats include advection, dispersion and diffusion along groundwater flows. Therefore, it is of great importance to accurately assess the hydrogeological and geochemical characteristics of the host rock because it acts as a flow medium. Normally, borehole investigations were used to evaluate the characteristics and the use of multi-packer system is more efficient and economical compared to standpipes, as it divides a single borehole into multiple sections by installing multiple packers. For effective analyses and groundwater sampling, the entire system is designed by preselecting sections where groundwater flow is clearly remarkable. The selection is based on the analyses of various borehole and rock core logging data. Generally, sections with a high frequency of joints and evident water flow are chosen. Analyzing the logging data, which can be considered continuous, gives several local points where the results exhibit significant local changes. These clear deviations can be considered outliers within the data set, and machine learning algorithms have been frequently applied to classify them. The algorithms applied in this study include DBSCAN (density based spatial clustering of application with noise), OCSVM (one class support vector method), KNN (K nearest neighbor), and isolation forest, of which are widely used in many applications. This paper aims to evaluate the applicability of the aforementioned four algorithms to the design of multi-packer system. The data used for this evaluation were obtained from DB-2 borehole logging data, which is a deep borehole locates near KURT.

PURPOSES : A model for minimizing cutting loss and determining the optimum layout of blocks in pavements was developed in this study. METHODS : Based on literature review, a model which included constraints such as the amount, volume, overlap, and pattern, was developed to minimize the cutting loss in an irregular pavement shape. The Stach bond, stretcher bond, and herringbone patterns were used in this model. The harmony search and particle swarm algorithms were then used to solve this model. RESULTS : Based on the results of the model and algorithms, the harmony search algorithm yielded better results because of its fast computation time. Moreover, compared to the sample pavement area, it reduced the cutting loss by 20.91%. CONCLUSIONS : The model and algorithms successfully optimized the layout of the pavement and they have potential applications in industries, such as tiling, panels, and textiles.

A causality exists between insect density and plant health, where plant health is affected by both the plant’s potential and environmental factors. In other words, causality is possible between insect density and environmental factors, allowing for the analysis of insect density based on these environmental factors. Machine learning enables studying insect density alongside environmental factors, providing insights into the causality between insects, the environment, and plant health. Machine learning is a methodology that involves the design of models by learning patterns from input data. This study aims to predict F. occidentalis density by sampling environmental factors and applying them to machine learning models.

Nowadays, artificial intelligence model approaches such as machine and deep learning have been widely used to predict variations of water quality in various freshwater bodies. In particular, many researchers have tried to predict the occurrence of cyanobacterial blooms in inland water, which pose a threat to human health and aquatic ecosystems. Therefore, the objective of this study were to: 1) review studies on the application of machine learning models for predicting the occurrence of cyanobacterial blooms and its metabolites and 2) prospect for future study on the prediction of cyanobacteria by machine learning models including deep learning. In this study, a systematic literature search and review were conducted using SCOPUS, which is Elsevier’s abstract and citation database. The key results showed that deep learning models were usually used to predict cyanobacterial cells, while machine learning models focused on predicting cyanobacterial metabolites such as concentrations of microcystin, geosmin, and 2-methylisoborneol (2-MIB) in reservoirs. There was a distinct difference in the use of input variables to predict cyanobacterial cells and metabolites. The application of deep learning models through the construction of big data may be encouraged to build accurate models to predict cyanobacterial metabolites.

Algorithms are rapidly altering the way society operates (SIOP, 2020). Algorithms are used in modern businesses for tasks such as hiring, advising investors on financial matters, recommending new products to customers (Shankar, 2018). However, lay people frequently oppose them, a phenomenon known as algorithm aversion (e.g., Dietvorst et al., 2015). While prior research tries to address this issue by identifying cognitive and affective predictors of algorithm aversion, we seek to contribute to the algorithm aversion literature by investigating an understudied antecedent of people’s support for algorithm adoption—their cultural values (Dietvorst & Bartels, 2022).

In this paper, we propose a method for diagnosing overload and working load of collaborative robots through performance analysis of machine learning algorithms. To this end, an experiment was conducted to perform pick & place operation while changing the payload weight of a cooperative robot with a payload capacity of 10 kg. In this experiment, motor torque, position, and speed data generated from the robot controller were collected, and as a result of t-test and f-test, different characteristics were found for each weight based on a payload of 10 kg. In addition, to predict overload and working load from the collected data, machine learning algorithms such as Neural Network, Decision Tree, Random Forest, and Gradient Boosting models were used for experiments. As a result of the experiment, the neural network with more than 99.6% of explanatory power showed the best performance in prediction and classification. The practical contribution of the proposed study is that it suggests a method to collect data required for analysis from the robot without attaching additional sensors to the collaborative robot and the usefulness of a machine learning algorithm for diagnosing robot overload and working load.

This paper proposes a model predictive controller of robot manipulators using a genetic algorithm to secure the best performance by performing parameter optimization with the genetic algorithm. Genetic algorithm is a natural evolutionary process modeled as a computer algorithm and has excellent performance in global optimization, so it is useful for tuning control parameters. The sliding mode controller and inverse dynamics controller are included in the lower part of the model prediction controller to minimize the problems caused by non-linearity and uncertainty of the robot manipulator. The performance superiority of the proposed method as described above has been confirmed in detail through a simulation study.

Most of the open-source decision tree algorithms are based on three splitting criteria (Entropy, Gini Index, and Gain Ratio). Therefore, the advantages and disadvantages of these three popular algorithms need to be studied more thoroughly. Comparisons of the three algorithms were mainly performed with respect to the predictive performance. In this work, we conducted a comparative experiment on the splitting criteria of three decision trees, focusing on their interpretability. Depth, homogeneity, coverage, lift, and stability were used as indicators for measuring interpretability. To measure the stability of decision trees, we present a measure of the stability of the root node and the stability of the dominating rules based on a measure of the similarity of trees. Based on 10 data collected from UCI and Kaggle, we compare the interpretability of DT (Decision Tree) algorithms based on three splitting criteria. The results show that the GR (Gain Ratio) branch-based DT algorithm performs well in terms of lift and homogeneity, while the GINI (Gini Index) and ENT (Entropy) branch-based DT algorithms performs well in terms of coverage. With respect to stability, considering both the similarity of the dominating rule or the similarity of the root node, the DT algorithm according to the ENT splitting criterion shows the best results.

최근의 기후변화는 연안에서 더욱 가속화되고 있어 연안에서의 해양 환경변화 감시의 중요성이 커지고 있다. 클로로필-a 농도는 해양 환경 변화의 중요한 지표 중 하나로 수십년 동안 여러 해색 위성을 통해 전구 해양 표층의 클로 로필-a 농도가 산출되었으며 다양한 연구 분야에 활용되었다. 하지만 연안 해역의 탁한 해수는 외해의 맑은 해수와는 구별되는 구성 성분과 광학적 특성으로 인해 나타나는 심각한 오차 때문에 일반적으로 사용되는 전지구 대양을 위하여 만들어진 클로로필-a 농도 알고리즘은 연안 해역에 대입할 수 없다. 또한 연안 해역은 해역에 따라 성분과 특성이 크게 달라져 통일된 하나의 알고리즘을 제시하기 어렵다. 이러한 문제점을 극복하기 위하여 연안의 탁도가 높은 해역에서는 구성 성분과 광학적 변동 특성을 고려한 다양한 알고리즘들이 개발되어 사용되어 왔다. 클로로필-a 농도 산출 알고리즘은 크게 경험적 알고리즘, 반해석적 알고리즘, 기계학습을 활용한 알고리즘 등으로 나눌 수 있다. 해수의 반사 스펙트 럼에 기반한 청색-녹색 밴드 비율이 기본적인 형태로 주로 사용된다. 반면 탁한 해수를 위해 개발된 알고리즘은 연안 해역에 존재하는 용존 유기물과 부유물의 영향을 상쇄시키기 위한 방식으로 녹색-적색 밴드 비율, 적색-근적외 밴드 비 율, 고유한 광학적 특성 등을 사용한다. 탁한 해수에서의 신뢰성 있는 위성 클로로필-a 농도 산출은 미래의 연안 해역을 관리하고 연안 생태 변화를 감시하는데 필수적이다. 따라서 본 연구는 탁도가 높은 Case 2 해수에서 활용되어온 알고리즘들을 요약하고, 한반도 주변해역의 모니터링과 연구에 대한 문제점을 제시한다. 또한 다분광 및 초분광 센서의 개발로 더욱 정확하고 다양한 해색 환경을 이해할 수 있는 미래의 해색 위성에 대한 발전 전망도 제시한다.

This study pursues to solve a batch of nonlinear parameter estimation (NPE) problems where a model interpreting the independent and the dependent variables is given and fixed but corresponding data sets vary. Specifically, we assume that the model does not have an explicit form and the discrepancy between a value from a data set and a corresponding value from the model is unknown. Due to the complexity of the problem, one may prefer to use heuristic algorithms rather than gradient-based algorithms, but the performance of the heuristic algorithms depends on their initial settings. In this study, we suggest two schemes to improve the performance of heuristic algorithms to solve the target problem. Most of all, we apply a Bayesian optimization to find the best parameters of the heuristic algorithm for solving the first NPE problem of the batch and apply the parameters of the heuristic algorithm for solving other NPE problems. Besides, we save a list of simulation outputs obtained from the Bayesian optimization and then use the list to construct the initial population set of the heuristic algorithm. The suggested schemes were tested in two simulation studies and were applied to a real example of measuring the critical dimensions of a 2-dimensional high-aspect-ratio structure of a wafer in semiconductor manufacturing.

In this study, a welding heat source model was presented and verified during fiber laser welding. The multi-layered heat source model is a model that can cover most of existing studies and can be defined with a simple formula. It consists of a total of 12 parameters, and an optimization algorithm was used to find them. As optimization algorithms, adaptive simulated annealing, multi island genetic algorithm, and Hooke-Jeeves technique were applied for comparative analysis. The parameters were found by comparing the temperature distribution when the STS304L was bead on plate welding and the temperature distribution derived through finite element analysis, and all three models were able to derive a model with similar trends. However, there was a deviation between parameters, which was attributed to the many variables. It is expected that a more clear welding heat source model can be derived in subsequent studies by giving a guide to the relational expression and range between variables and increasing the temperature measurement point, which is the target value.