Dynamic Evaluation Method of the EW-AHP Attribute Identification Model for the Tunnel Gushing Water Disaster under Interval Conditions and Applications

The gushing water disaster in tunnels is a kind of harmful and risky engineering disaster. It has become a key problem to evaluate the risk of tunnel gushing water accurately and objectively. A case study of a typical highway tunnel is performed for theory and practice analysis. For this reason, the risk identification is carried out on the assessed objects, and 10 evaluation indexes are determined. In turn, the risk evaluation index system and classification standard are established. Furthermore, the entropy weight method and the analytic hierarchy process are combined to assign the weight to each evaluation index. Therefore, a dynamic risk assessment system, including the pre-evaluation model and the postevaluation model, is constructed with the attribute identification model. As a result, the tunnel section with a high risk of water inrush is accurately assessed, which is consistent with the construction situation on site. Moreover, it is verified that the assessment results are reliable, which can provide a reference for the similar projects.

Effect of Product Involvement on Panels’ Vocabulary Generation, Attribute Identification, and Sample Configurations in Beer

The aim of this study was to compare the performance of two semi-trained panels with different degrees of self-reported beer involvement in terms of beer consumption pattern. The two panels were beer non-drinkers (indicating willingness to taste beer) and craft-style beer drinkers. Eleven modified beer samples were evaluated during three separate tasks by both panels. The tasks were (1) a vocabulary generation on a sample level, (2) an attribute identification task with a list of attributes to choose from, and (3) a descriptive analysis. The performance of the two panels was evaluated and compared using three parameters, as follows: Descriptive similarity, attribute knowledge similarity, and perceptual similarity. The results showed that the craft-style beer drinkers generated the most precise vocabulary and correctly identified more attributes, compared to the beer non-drinkers. Furthermore, the sample sensory spaces generated by the two panels were different before the training period, but were perceptually similar post training. To conclude, the beer consumption pattern influenced all aspects of panel performance before training, with the craft-style panel performing better than the non-drinkers panel. However, the panels’ performance became more similar after a short period of training sessions.

TallyQA: Answering Complex Counting Questions

Most counting questions in visual question answering (VQA) datasets are simple and require no more than object detection. Here, we study algorithms for complex counting questions that involve relationships between objects, attribute identification, reasoning, and more. To do this, we created TallyQA, the world’s largest dataset for open-ended counting. We propose a new algorithm for counting that uses relation networks with region proposals. Our method lets relation networks be efficiently used with high-resolution imagery. It yields stateof-the-art results compared to baseline and recent systems on both TallyQA and the HowMany-QA benchmark.