There exists a great demand for automatic anomaly detection in industrial world. The anomaly has been defined as a group of samples that rarely or never appears. Given a type of products, one has to collect numerous samples and train an anomaly detector. When one diverts a model trained with old types of products with sufficient inventory to the new type, one can detect anomalies of the new type before a production line is established. However, because of the definition of the anomaly, a typical anomaly detector considers the new type of products anomalous even if it is consistent with the standard. Given the above practical demand, this study propose a novel problem setting, few-shot anomaly detection, where an anomaly detector trained in source domains is adapted to a small set of target samples without full retraining. Then, we tackle this problem using a hierarchical probabilistic model based on deep learning. Our empirical results on toy and real-world datasets demonstrate that the proposed model detects anomalies in a small set of target samples successfully.

Recently, the environmental impact has been required to be explicitly taken into account in the life cycle design of building structures besides the initial economical efficiency which has been ordinarily required up to the present days. Consequently, it is important not only to design the building structures so as to be safe as well as economical at the initial stage of completion of the building but also to pay attention and take into account the total behaviors of the designed and constructed building structures during all stages of their life cycles. The former paper by the authors has proposed the way to deal with the life cycle design of building structures with consideration of a hierarchy according to the ranks of building construction systems by using genetic algorithms as the optimization scheme with respect to both ecology and economy for evaluations. Authors have been developing computational software for the life-cycle design of structures through utilization of Genetic Algorithm. In this paper, the presumption method for the lifetime of structures is introduced, by which easy usage as well as practicality of the tool are expected. For evaluation of the lifetime of structures, the lifetime of each portion is individually evaluated where surrounding situations of them, such as near from water, exposed to outside of building and so on, are taken into consideration. This newly introduced treatment is expected to exclude the arbitrariness on the evaluation time of the life-cycle design and make the tool much more suitable for practical use.

No abstract

No abstract

No abstract