Collaborative Metric Learning (CML) is a recommendation model based on an embedded representation trained by implicit data, i.e., the behavioral histories of clicking and browsing. CML learns a metric space to embed users and items considering not only the relationship between users and items, but also item–item similarity and user–user similarity. Moreover, CML can recommend items close to each user in the trained space that match the preferences of the user. However, CML tends to be influenced strongly by popular items among many users, and the accuracy of embedded representations of other minor items is often neglected. It is necessary to learn the embedded representations of minor items that match user preferences with higher accuracy to provide unexpected recommendations of items that users may not be aware of in advance. In this article, a method is proposed to learn the embedded representations that capture user preferences by weighting loss functions according to the number of observations of implicit data and to make unexpected and effective recommendations, including minor items. Finally, the proposed method is applied to an actual movie evaluation dataset, and the usefulness of the proposed method in making unexpected recommendations based on user preferences is demonstrated.
The inspection process, which was previously based primarily on human sampling, has now been fully automated by the introduction of inspection machines that use images. However, depending on the inspection conditions of the system, there is a risk of the over- or non-detection of defective products. This problem may result in a loss for companies that have implemented automated inspection machines. Therefore, with the introduction of such machines, the establishment of discrimination techniques for defect detection has become a critical issue. This study aimed to develop a method for automatically detecting defects in image data with high accuracy. This study used images of metal products. We attempted to detect defects inside metal products. This study proposed a method using machine learning. Additionally, this research made recommendations for data preprocessing that may be required before machine learning. Then, we chose the best combination of the proposed machine learning approach and data preprocessing, that is the most capable of detecting defects, using the L16 orthogonal table or ANOVA. The classification accuracy exceeded 90% using the proposed method, and we partly solved the problem thrown up by previous studies.
Condition-based maintenance of systems that deteriorates in accordance with a semi-Markov process was investigated. In addition to considering the system deterioration state as a criterion for maintenance decision-making, the cumulative damage caused by random external shocks was considered as well. The occurrence of random shocks was assumed to follow a Poisson process. Furthermore, the damage from random external shocks was assumed to be affected by the system deterioration state, and the system deterioration process was assumed to be affected by the cumulative damage from random external shocks. The system operator chooses the maintenance actions (continue to operate, repair, or replace) that minimize the total expected discounted cost over an infinite time horizon at the time when the system deterioration state changes. The maintenance policy optimization problem is formulated as a semi-Markov decision process considering cumulative damage from random external shocks. The properties of the total expected discounted cost with respect to the system deterioration state and cumulative damage are proven to be monotone. A threshold-type maintenance policy is derived on the basis of the monotone properties. Numerical examples illustrate the threshold-type maintenance policy.
Many companies, including food manufacturers, are developing and manufacturing environment-friendly containers and packaging in response to increasingly serious environmental problems and to achieve sustainable development goals. For consumers to adopt environmentally conscious behavior when purchasing, using, and disposing of containers and packaging, they need to correctly understand the environmentally conscious efforts of companies and product information. They must also dispose of and recycle products according to the rules of the municipality in which they reside.
Companies and consumers may differ in their environment-friendly efforts in terms of their ideal state and current conditions, and their understanding of environmentally conscious behavior may also differ. This study defines these differences as gaps and aims to identify them by surveying consumers on their environmental awareness and behavior. First, five hypotheses for these gaps were formulated based on interviews with consumers and companies, and a survey was designed to test them. Second, survey data were analyzed to test these hypotheses. For example, the surveys revealed that consumers perceived measures that are easy to understand as being environment-friendly, such as switching from plastic to paper, to be effective.
In recent years, marketing has often relied on using attribute information associated with the accounts registered in online services. However, most people use such services without a personal account, and it is impossible to obtain attribute information for these unregistered users. To deal with this situation, semi-supervised learning is an effective way to increase the number of users with attribute information. Attributes can be predicted from the historical data of unregistered users, having the historical data of registered users who have attribute information. One such semi-supervised learning method is the ladder network, which is a neural-network-based model that adds and removes noise. This model provided highly accurate predictions for image data and is also considered to be useful for predicting user attributes from historical data, where the feature vector is high-dimensional. However, this method does not account for cases where the label takes an ordered value, such as the user’s age category. In this study, we propose an extended model based on a ladder network that incorporates a mechanism that can appropriately predict a user’s attribute information, including ordinal-scale variables. We also conducted an evaluation experiment using browsing history data to demonstrate the effectiveness of the proposed method.