Multi-modal pre-training models have been intensively explored to bridge vision and language in recent years. However, most of them explicitly model the cross-modal interaction between image-text pairs, by assuming that there exists strong semantic correlation between the text and image modalities. Since this strong assumption is often invalid in real-world scenarios, we choose to implicitly model the cross-modal correlation for large-scale multi-modal pretraining, which is the focus of the Chinese project 'Wen-Lan' led by our team. Specifically, with the weak correlation assumption over image-text pairs, we propose a twotower pre-training model called BriVL within the crossmodal contrastive learning framework. Unlike OpenAI CLIP that adopts a simple contrastive learning method, we devise a more advanced algorithm by adapting the latest method MoCo into the cross-modal scenario. By building a large queue-based dictionary, our BriVL can incorporate more negative samples in limited GPU resources. We further construct a large Chinese multi-source imagetext dataset called RUC-CAS-WenLan for pre-training our BriVL model. Extensive experiments demonstrate that the pre-trained BriVL model outperforms both UNITER and OpenAI CLIP on various downstream tasks.

No abstract

Abstract:The reasonable utilization of limited resources is critical to realize the sustainable developments. In the initial 72-h crucial rescue period after the disaster, emergency supplies have always been insufficient and the demands in the affected area have always been uncertain. In order to improve timeliness, utilization and sustainability of emergency service, the allocation of the emergency supplies and the emergency vehicle routes should be determined simultaneously. Assuming the uncertain demands follow normal distribution, an optimization model for the emergency vehicle routing, by considering the insufficient supplies and the uncertain demands, is developed. The objective function is applied to minimize the total costs, including the penalty costs induced by more or less supplies than the actual demands at all demand points, as well as the constraints of the time windows and vehicle load capacity taken into account. In more details, a solution method for the model, based on the genetic algorithm, is proposed, which solves the problem in two stages. A numerical example is presented to demonstrate the efficiency and validity of the proposed model and algorithm.

Based on the single pricing method of the high-speed railway (HSR) in China, a pricing strategy without flexibility leads to the problem of extreme fluctuations in passenger flow and difficulty in increasing revenue. In order to achieve sustainable development of the HSR from the perspective of pricing, in this study, we divided the passenger market according to the different factors affecting passengers' choice behavior, maximized ticket sales revenue with expected travel cost as the reference point, and used prospect theory to construct a differentiated pricing model under elastic demand. A simulated annealing algorithm was used to solve this model under two passenger flow intensities. Taking the Beijing-Shanghai corridor as an example for analysis, the results show that differential pricing can be implemented on the basis of passenger decision-making, and price reductions at off-peak periods will attract passenger flow which will increase ticket sales revenue by 10.41%. During the peak period, prices can be increased to maintain passenger flow, and ticket sales revenue will increase by 7.98%. We also found that increasing passenger expectations have a greater impact on ticket sales. This study provides theoretical and methodological support for the sustainable development of the HSR.

This research investigates the effect of fairness concerns on a sustainable low-carbon supply chain (LCSC) with a carbon quota policy, in which a manufacturer is in charge of manufacturing low-carbon products and sells them to a retailer. The demand is affected by price and the carbon emission reduction rate. The optimal decisions of pricing and carbon emission reduction rate are analyzed under four decision models: (i) centralized decision, (ii) decentralized decision without fairness concern, (iii) decentralized decision with manufacturer’s fairness concern, (iv) decentralized decision with retailer’s fairness concern. The results indicate that the profits in the centralized LCSC are higher than those in the decentralized LCSC with fairness concern. If a manufacturer pays close attention to fairness, the fairness concern coefficient will reduce the carbon emission reduction rate and the profit of the LCSC and increase the wholesale price and the retail price of the product. If a retailer pays close attention to fairness, and the preference of consumers for a low-carbon product is low, the fairness concern coefficient of the retailer increases the total profit of the LCSC and decreases the carbon emission reduction rate and retail price of the product. Otherwise, if the preference of consumers for a low-carbon product is great, the fairness concern coefficient of the retailer would lead to a lower retail price compared with the retail price in the centralized decision and decrease the total profit of the LCSC.