一种基于感受野增强的人脸检测方法

为了解决多任务级联卷积神经网络（MTCNN）算法网络模型在小人脸检测方面鲁棒性较低的问题,提出了一种基于感受野增强的网络模型。首先,为MTCNN算法模型中的R-Net网络和O-Net网络添加感受野模块（receptive field blocks,RFB-S）。其次,通过添加批量标准化和全局平均池化,加速网络模型的收敛,防止模型过拟合。最后,调整网络任务的权重,P-Net和R-Net网络用于人脸区域粗筛选,O-Net网络用于人脸区域精筛选以及人脸关键点回归。实验结果表明,与MTCNN算法网络模型相比,所提模型缩小了16%,但检测速度提升了9%,在FDDB数据集上的检测精度提高了2.3%。因此,基于感受野增强的网络模型能有效完成人脸的检测任务,增强对小人脸检测的鲁棒性,可为人脸识别、表情识别等提供技术支持。     

Combining value and momentum: evidence from the Nordic equity market

This is the first article that explores the recently proposed average ranking approach for the value and momentum strategy in the Nordic equity market offering an exceptional experimental environment. Our results indicate that in the Nordic stock markets, the value anomaly offered excess returns in the 1993–2017 sample period only when small stocks were a part of the portfolio, whereas the momentum effect is strong and significant, irrespective of size. Interestingly, our findings also indicate that the negative correlation between value and momentum seems to be driven by growth stocks: Winner stocks that are value stocks generated 1.66% per month on average, whereas winner stocks that are growth stocks exhibit virtually the same average payoff. On the other hand, the spread between value and growth stocks that are loser stocks is on average 0.97% per month.     

A Dynamic Fuzzy Money Management Approach for Controlling the Intraday Risk-Adjusted Performance of AI Trading Algorithms

The majority of existing artificial intelligence (AI) studies in computational finance literature are devoted solely to predicting market movements. In this paper we shift the attention to how AI can be applied to control risk-based money management decisions. We propose an innovative fuzzy logic approach which identifies and categorizes technical rules performance across different regions in the trend and volatility space. The model dynamically prioritizes higher performing regions at an intraday level and adapts money management policies with the objective to maximize global risk-adjusted performance. By adopting a hybrid method in conjunction with a popular neural network (NN) trend prediction model, our results show significant performance improvements compared with both standard NN and buy-and-hold approaches. Copyright © 2014 John Wiley & Sons, Ltd.     

The Impact of Better‐than‐Average Bias and Relative Performance Pay on Performance Outcome Satisfaction

Drawing on equity and expectancy theories, we hypothesize that the perception of accountants about their ability to contribute relative to a peer (operationalized as the better‐than‐average [BTA] bias) negatively influences their satisfaction with the outcomes of the performance evaluation process (operationalized as performance outcome satisfaction [POS]). We hypothesize further that this negative influence is mitigated by the amount of relative performance pay. We test these hypotheses using data collected from a survey of and an experiment involving 164 entry‐level accountants. We found that in general our participants rated themselves better than the average audit professional and their immediate work associate; that is, they displayed a BTA bias. Moreover, we found that both the BTA bias and performance pay individually influenced POS; we also found a moderately significant interaction effect. In their entirety, the results indicate that the greater an entry‐level accountant believes that she or he is better than average the more likely her or his performance outcome satisfaction will fall.     

‘Street Music’, Urban Ethnography and Ghettoized Communities

Hip‐hop is a definitively urban movement, born in the crisis of the Fordist city and rooted in the 1970s street culture of poor and working‐class African Americans and Latinos in New York City. Engaging with the contributions of Beer and Lamotte, this essay addresses two questions. Firstly, can we understand hip‐hop as a politics of resistance, a social movement rooted in a claim on urban space and a practice of urban citizenship? And secondly, is hip hop, and particularly rap music, a form of urban and regional research? I argue that as a primarily artistic movement and black expressive culture (subject to commercial imperatives), hip‐hop has a layered and complex relationship with the social, political and spatial fabric of urban America. The complexity of this relationship renders problematic attempts to understand hip‐hop as urban ethnography or as a resistance politics. I conclude by discussing the potential of an engagement between hip‐hop and critical urban studies.     

Technically and cost-efficient centralized allocations in data envelopment analysis

In data envelopment analysis and with a variable returns to scale production-technology, we apply Banker's [2] approach to determine the relationship between technically and cost-efficient industry structures, featuring reallocation of outputs and a variable number of firms. The interpretation based on the most productive and optimal scale-size notions allows us to both establish an inequality relationship between the corresponding industry-efficiency measures and provide adequate information on optimal solutions. At the applicative level, we introduce an exact algorithm to solve related non-linear programming problems, thus providing the decision maker with an accurate method for computing and comparing the input and output mixes and the optimal number of units obtained in the two allocations. Empirical illustration, given with reference to the Italian local-public-transit sector and employing a multiple inputs and outputs technology, reveals striking differences with regard to the managerial and regulatory implications of the two centralized allocations.     

中国市场投资的问题与对策

任何投资的目的都是要实现利润的最大化,既使企业的价值或股东的财富最大化。在市场条件下,有很多不可知的变量,这些变量就造成了投资过程中的风险。本文就目前市场投资的热点问题:如何认识投资并确定投资对象,在分析项目可行性的基础上进行了探讨。     

导线测量与水准测量外业记录的技巧

在小区域控制测量中,导线测量和水准测量是应用最普遍的方法,提高其外业数据的记录效率与正确率是非常有意义的。本文介绍了导线测量中的测回法与三、四等水准测量的快速记录技巧,通过使用这些技巧,可以提高外业记录工作的效率,降低外业工作的强度。     

Computational aspects of robust optimized certainty equivalents and option pricing

Accounting for model uncertainty in risk management and option pricing leads to infinite‐dimensional optimization problems that are both analytically and numerically intractable. In this article, we study when this hurdle can be overcome for the so‐called optimized certainty equivalent (OCE) risk measure—including the average value‐at‐risk as a special case. First, we focus on the case where the uncertainty is modeled by a nonlinear expectation that penalizes distributions that are “far” in terms of optimal‐transport distance (e.g. Wasserstein distance) from a given baseline distribution. It turns out that the computation of the robust OCE reduces to a finite‐dimensional problem, which in some cases can even be solved explicitly. This principle also applies to the shortfall risk measure as well as for the pricing of European options. Further, we derive convex dual representations of the robust OCE for measurable claims without any assumptions on the set of distributions. Finally, we give conditions on the latter set under which the robust average value‐at‐risk is a tail risk measure.