No abstract

No abstract

Transportation accounts for more than a quarter of the greenhouse gas emissions that are causing climate change. Carpooling is a subset of the sharing economy, in which individuals share their vehicle with commuters to save travel expenses. In recent decades, carpooling has been promoted as a feasible alternative to car ownership with the potential to alleviate traffic congestion, parking demand, and environmental problems. Unstable economic conditions, cultural norms, and lack of infrastructure make cultural exchange activities and mobility habits different in developing nations to those in developed countries. The rapid evolution of sharing mobility has reshaped travelers’ behavior and created a dire need to determine the travel patterns of commuters living in megacities in developing countries. To obtain data, a web-based stated choice (SC) experiment was used in this study. It used mode-related variables, socioeconomic demographic variables, and a coronavirus disease 2019 (COVID-19) precautionary measure variable. Logit models, namely the mixed logit regression model (ML) and the multinomial logit regression model (MNL), were applied to analyze the available data. According to modeling and survey data, economic variables associated with modes of transport, such as trip time and trip cost, were determined to be significant. Additionally, the results revealed that commuters were more conscious of COVID-19 preventive measures, which was determined to be highly significant. The findings showed that the majority of residents in the COVID-19 pandemic continue to rely on automobiles and motorcycles. It is noteworthy that individuals with more than two members in their family and a travel distance of less than seven miles were more likely to prefer a carpooling service. This study’s findings will provide a basis for researchers to aid existing operators in the field of transportation, as well as offer guidelines for governments in developing countries to enhance the utility of transportation networks.

No abstract

This article mainly discusses how to extract the interested information from massive amounts of micro-blogs and recommend right information to user, which is a hot research area in recommendation systems and social networks, too. To solve this problem, a model called Multi-tags Latent Dirichlet Allocation is proposed. Using this model, topics paid attention by users can be mined effectively and the defect of low degree of differentiation for the short blog content is settled. Experiments showed that the tags of user's micro-blog can be figured out with this model which makes users manage their resources at their convenience and others find their needed resources through tags. The results, experimented on real micro-blog data set, indicate that this model works better than traditional model on extracting tags. Standard measuring index Perplexity is applied to this model to estimate the likelihood of new text. If the number of topics is selected appropriately, the accuracy will be raised to almost 10%.

Transfer learning is a problem defined over two domains. These two domains share the same feature space and class label space, but have significantly different distributions. One domain has sufficient labels, named as source domain, and the other domain has few labels, named as target domain. The problem is to learn a effective classifier for the target domain. In this paper, we propose a novel transfer learning method for this problem by learning a partially shared classifier for the target domain, and weighting the source domain data points. We learn some shared subspaces for both the data points of the two domains, and a shared classifier in the shared subspaces. We hope that in the shared subspaces, the distributions of two domain can match each other well, and to match the distributions, we weight the source domain data points with different weighting factors. Moreover, we adapt the shared classifier to each domain by learning different adaptation functions. To learn the subspace transformation matrices, the classifier parameters, and the adaptation parameters, we build a objective function with weighted classification errors, parameter regularization, local reconstruction regularization, and distribution matching. This objective function is minimized by an iterative algorithm. Experiments show its effectiveness over benchmark data sets, including travel destination review data set, face expression data set, spam email data set, etc.

No abstract

Abstract:Image mosaic is an important research content in digital image processing, and could be used to solve the problem of observing large objects with narrow view, such as 360 degree panorama stitching. Microscopic observation of liver biopsy is a commonly used method in diagnosing liver diseases, which are always rely on the whole liver slice. Therefore, the liver pathological microscopic image mosaic is the best way to formulate it. This paper proposes a liver image mosaicing system based on scale invariant feature transform and point set matching method, which includes the feature point selection and location process to find the extremum point, screen them, and precisely position them. This system takes the affine transformation as the motion model, and adopts a new matching algorithm for the rigid transformation to accelerate solving the motion parameter. The design and implementation of the liver image mosaicing system have completed more than 100 cases successfully, which shows the effectiveness and robustness of our algorithms.