Case studies of Artificial Intelligence, Business Intelligence, Analytics Technologies for Industry Platforms
Permanent URI for this collection
1 - 6 of 6
ItemUtilizing Active Machine Learning for Quality Assurance: A Case Study of Virtual Car Renderings in the Automotive Industry( 2022-01-04)Computer-generated imagery of car models has become an indispensable part of car manufacturers' advertisement concepts. They are for instance used in car configurators to offer customers the possibility to configure their car online according to their personal preferences. However, human-led quality assurance faces the challenge to keep up with high-volume visual inspections due to the car models’ increasing complexity. Even though the application of machine learning to many visual inspection tasks has demonstrated great success, its need for large labeled data sets remains a central barrier to using such systems in practice. In this paper, we propose an active machine learning-based quality assurance system that requires significantly fewer labeled instances to identify defective virtual car renderings without compromising performance. By employing our system at a German automotive manufacturer, start-up difficulties can be overcome, the inspection process efficiency can be increased, and thus economic advantages can be realized.
ItemThe Role of Technical and Process Quality of Chatbots: A Case Study from the Insurance Industry( 2022-01-04)Artificial intelligence (AI) influences customer service through benefits, such as reliability, availability, and economic efficiency. However, AI applications also involve challenges of user acceptance and quality concerns. To address these challenges, we investigate the factors that impact AI preference and adoption among users of a chatbot in a real customer service scenario. We focus on Emma, a customer service chatbot at a large Finnish insurance company. Our analysis, based on an online survey administered to 225 consumers using the chatbot for their customer service needs, shows that customers are reasonably satisfied with Emma, though they are generally do not prefer AI over a human. Users’ perceived process quality relating to “soft” aspects of interaction is quintessential in strengthening technical quality relating to effectiveness and efficiency of service, both contributing to AI preference. Thus, the chatbot’s problem-solving ability acts as a hygiene factor, which alone cannot ensure adoption. As a pleasing and useful interaction is prerequisite for user experience, organizations should consider both technical and process quality when implementing chatbots in customer service.
ItemAutomated Defect Detection of Screws in the Manufacturing Industry Using Convolutional Neural Networks( 2022-01-04)Defect detection in industrial production processes is an important and necessary part of quality control. Many defects can occur during the manufacturing process, causing high manufacturing costs. Thus the inspection of screws, which represent an indispensable element of many mechanical components, is a critical process. To reduce manufacturing costs and increase efficiency, a reliable method for inspection is Deep Learning. It can help simplify the process of quality control and increase the velocity and volume of detected defects in screws. This approach uses a CNN model to classify non-defective and defective screws with different types of defects. Instead of manual quality control methods, which can be easily biased, our CNN approach is accurate, cost-efficient, and fast, with an accuracy of over 97 percent. With this approach corresponding to industrial production processes, different defects in screws and non-defective screws can be classified from images according to a real-world industrial inspection scenario.
ItemAn Innovative Approach to Modeling Aviation Safety Incidents( 2022-01-04)Due to the complexity of aviation safety operations, the number of flight incidents continues to rise. The Aviation Safety Reporting System (ASRS) contains the largest collection of such incidents. Efficient and effective analysis of these incidents remains a challenge. This paper proposes a new approach to analyze aviation safety records using deep learning methods to improve incident classification. The proposed approach, CNN-LSTM, combines the characteristics of convolutional neural network (CNN) and long short-term memory (LSTM) neural network, and a distributed computing method to model aviation safety data. The five machine learning methods Logistic Regression, Naive Bayes, Random Forest, Support Vector Machine, Multi-layer Perceptron were used to compare with CNN-LSTM. The results show that CNN-LSTM model can significantly improve the accuracy rates of classification for aviation safety incident reports using Word2Vec. The distributed platform in Spark with clusters can make full use of computing resources when processing textual data from ASRS, reducing time-consumption greatly when compared with machine learning algorithms running on a standalone computer. Timely and accurate identification of causes of reported incidents is important. The results of this study demonstrate a new approach to improve both accuracy and efficiency in incident cause identification.
ItemAn Empirical Study of Factors Affecting Language-Independent Models( 2022-01-04)Scaling existing applications and solutions to multiple human languages has traditionally proven to be difficult, mainly due to the language-dependent nature of preprocessing and feature engineering techniques employed in traditional approaches. In this work, we empirically investigate the factors affecting language-independent models built with multilingual representations, including task type, language set and data resource. On two most representative Natural Language Processing tasks --- sentence classification and sequence labeling, we show that language-independent models can be comparable to or even outperforms the models trained using monolingual data, and they are generally more effective on sentence classification. We experiment language-independent models with many different languages and show that they are more suitable for typologically similar languages. We also explore the effects of different data sizes when training and testing language-independent models, and demonstrate that they are not only suitable for high-resource languages, but also very effective in low-resource languages.