基于多尺度注意力半监督学习的老照片划痕修复

老照片由于长时间的磨损或保存不当,会出现照片的划痕损伤。随着深度学习在图像重建中的应用,基于深度学习方法能够在纹理修复的基础上获取图像的语义信息并预测语义内容,使老照片修复的整体效果更加符合客观事实,但利用深度学习进行老照片划痕修复缺乏学习所需数据集。提出一种基于半监督学习的老照片划痕自动修复的方法,创建划痕合成数据集SynOld用于网络训练,同时搜集真实的划痕老照片用于训练和测试,将合成数据集和真实老照片加入网络学习,两者共享网络参数,并通过鉴别器来区分网络生成图像与真实图像。对于合成数据集有监督的分支采用均方差损失、感知损失和对抗损失约束训练,对于真实老照片无监督的分支采用总变差损失控制训练。实验结果表明,相比于多尺度特征注意力网络的监督学习方法,该方法在合成数据集SynOld和真实老照片上都具有较好的修复效果。     

An option pricing by eliminating observation noise when the model parameters are unknown

The stock price that is observed in the market contains irregular fluctuations that cannot be explained by the theory. We assume this kind of fluctuation to be observation noise. By eliminating observation noise from the underlying stock price, we can evaluate the option that better reflects the entity price. We define the stock price that eliminates observation noise as the entity stock price. The most important problem in eliminating the observation noise is the choice of unknown constant parameters of the market model. We develop a parameter estimation method by using the relationship between stock price and its option price or future price. We obtain satisfactory results in testing the algorithm by some simulations. We apply the parameters obtained by this method to estimate the stock price by eliminating observation noise from the real market data. We then calculate the call option price based on the estimated entity stock price.     

Agent-based joint model of residential location choice and real estate price for land use and transport model

Residential location choice (RLC) and real estate price (REP) models are traditional and key components of land use and transport model. In this study, an agent-based joint model of RLC and REP (RLC–REP model) was proposed for SelfSim, an agent-based dynamic evolution of land use and transport model. The RLC–REP model is capable of simulating the negotiation between the active household agents (buyers) and owner agents (sellers) using agent-based modeling. In particular, both utility maximization theory and prospect theory were used to develop a utility function to simulate the location choice behavior of active household agents. The utility function incorporates only two variables: house price and accessibility. The latter variable is calculated using MATSim, an activity-based model. The asking price behavior of owner agents is based on three specific rules. The residential location choices of household agents and house prices can be obtained by negotiation. Finally, genetic algorithm was used to estimate the parameters of the RLC–REP model. The calibrated model was tested in Baoding, a medium-sized city in China, and historical validation was performed to assess its performance. The results suggest that the forecasting ability of the RLC–REP model in terms of real estate price is satisfactory.     

基于ARIMA模型的二手房价格预测

 摘要： 近年来，我国一二线城市房价持续上涨，房屋成了人们日常生活讨论的热门话题，大家纷纷对未来的房价走势做出猜测。本文爬取国内某知名大型房产网站自2013年以来广州和深圳的二手房均价数据，采用ARIMA模型对未来的房价进行滚动预测，并使用RMSE对预测精度进行判断。结果表明，该模型可以对二手房均价进行持续预测，且预测精度较高，可为房屋买卖者提供参考。     

房地产交易核价系统价格评估方法分析

对目前我国在房地产交易价格评估时常用的三种评估方法进行了分析研究,提出了一种基于市场法的房地产交易价格评估循环算法：该算法计算结果可靠,已应用于房地产交易税收核价系统中。     

Estimation of Bid Functions for Location Choice and Price Modeling with a Latent Variable Approach

A new approach for the estimation of bid-rent functions for residential location choice is proposed. The method is based on the bid-auction approach and considers that the expected maximum bid of the auction is a latent variable that can be related to observed price indicators through a measurement equation. The method has the advantage of allowing for the estimation of the parameters of the bid function that explain the heterogeneous preferences of households for location while simultaneously adjusting the expected maximum bid to reproduce realistic values. The model is applied and validated for a case study on the city of Brussels. Results show that the proposed model outperforms other methods for bid-rent estimation, both in terms of real estate prices and spatial distribution of agents, especially when detailed data describing the real estate goods and their prices is not available.     

基于二元语义多属性决策方法的购房评估研究

提出了一种基于二元语义的购房评估方法。在经过调查分析基础上,构建了购房评估模型,将二元语义多属性决策方法应用于购房评估中,为购房者提供参考决策。最后通过实例说明了该方法运用于购房评估的具体过程,并验证了其合理性和可操作性。     

House Price Appreciation in Old Age: Analysis and Issues for the Use of Reverse Mortgages as a Retirement Funding Strategy in Australia

This paper investigates whether the houses of Australian elderly home owners appreciate at below the market rate and examines the issues this may raise for the use of reverse mortgages as a retirement funding strategy in Australia. The viability of reverse mortgages where elderly home owners effectively borrow against their housing equity depends strongly on house prices appreciating enough to offset the outstanding loan balance at the end of the loan tenure. This paper’s findings indicate that after controlling for other influences, being aged 75 years or over lowers annual house price appreciation rate by almost 1.4 percentage points. Being aged 75 years or over also lowers home improvement expenditure by over AUD3,000 per year and this is found to be attributable to a decline in income during old age. The majority of elderly home owners want to protect at least half of their housing equity when considering participating in reverse mortgage programs, but given below-average house price appreciation rates during old age, the propensity of a 50% equity protection declines sharply with age. In particular, single females aged 75 years or over are least able to protect at least half of their housing equity, with only around 15% able to do so by the end of a reverse mortgage loan tenure. The paper also finds that, worryingly, elderly home owners with characteristics associated with slower house price appreciation rates are over-represented among reverse mortgage borrowers in Australia, namely, those aged 75 years or over, single, living in apartments or residing in states with relatively slow house price growth.     

小波神经网络在房地产价格指数预测中的应用

随着房地产价格指数的作用充分显现,探求预测房地产价格指数的有效方法是需深入研究的方向。该文以中房上海住宅价格指数为例,首先对房地产价格指数序列性质进行分析,表明房地产价格指数是具有非线性特征的非平稳时间序列。采用小波神经网络对房地产价格指数进行预测,并将预测结果与指数平滑法和RBF神经网络预测做了对比。采用MATLAB对拟合和预测过程进行仿真。结果指标表明,在大样本数据的情况下,采用小波神经网络对房地产指数进行预测能够获得较好的效果。     

基于集成LightGBM和贝叶斯优化策略的房价智能评估模型

针对传统房价评估方法中存在的数据源单一、过分依赖主观经验、考虑因素理想化等问题,提出一种基于多源数据和集成学习的智能评估方法。首先,从多源数据中构造特征集,并利用Pearson相关系数与序列前向选择法提取最优特征子集;然后,基于构造的特征,以Bagging集成策略作为结合方法集成多个轻量级梯度提升机（LightGBM）,并利用贝叶斯优化算法对模型进行优化;最后,将该方法应用于房价评估问题,实现房价的智能评估。在真实的房价数据集上进行的实验表明,相较于支持向量机（SVM）、随机森林等传统模型,引入集成学习和贝叶斯优化的新模型的评估精度提升了3.15%,并且百分误差在10%以内的评估结果占比84.09%。说明所提模型能够很好地应用于房价评估领域,得到的评估结果更准确。     

基于集成LightGBM和贝叶斯优化策略的房价智能评估模型

针对传统房价评估方法中存在的数据源单一、过分依赖主观经验、考虑因素理想化等问题，提出一种基于多源数据和集成学习的智能评估方法。首先，从多源数据中构造特征集，并利用Pearson相关系数与序列前向选择法提取最优特征子集；然后，基于构造的特征，以Bagging集成策略作为结合方法集成多个轻量级梯度提升机（LightGBM），并利用贝叶斯优化算法对模型进行优化；最后，将该方法应用于房价评估问题，实现房价的智能评估。在真实的房价数据集上进行的实验表明，相较于支持向量机（SVM）、随机森林等传统模型，引入集成学习和贝叶斯优化的新模型的评估精度提升了3.15%，并且百分误差在10%以内的评估结果占比84.09%。说明所提模型能够很好地应用于房价评估领域，得到的评估结果更准确。     

Flood depth mapping in street photos with image processing and deep neural networks

Many parts of the world experience severe episodes of flooding every year. In addition to the high cost of mitigation and damage to property, floods make roads impassable and hamper community evacuation, movement of goods and services, and rescue missions. Knowing the depth of floodwater is critical to the success of response and recovery operations that follow. However, flood mapping especially in urban areas using traditional methods such as remote sensing and digital elevation models (DEMs) yields large errors due to reshaped surface topography and microtopographic variations combined with vegetation bias. This paper presents a deep neural network approach to detect submerged stop signs in photos taken from flooded roads and intersections, coupled with Canny edge detection and probabilistic Hough transform to calculate pole length and estimate floodwater depth. Additionally, a tilt correction technique is implemented to address the problem of sideways tilt in visual analysis of submerged stop signs. An in-house dataset, named BluPix 2020.1 consisting of paired web-mined photos of submerged stop signs across 10 FEMA regions (for U.S. locations) and Canada is used to evaluate the models. Overall, pole length is estimated with an RMSE of 17.43 and 8.61 in. in pre- and post-flood photos, respectively, leading to a mean absolute error of 12.63 in. in floodwater depth estimation. Findings of this research are sought to equip jurisdictions, local governments, and citizens in flood-prone regions with a simple, reliable, and scalable solution that can provide (near-) real time estimation of floodwater depth in their surroundings.     

Quasi-Monte Carlo estimation in generalized linear mixed models

Generalized linear mixed models (GLMMs) are useful for modelling longitudinal and clustered data, but parameter estimation is very challenging because the likelihood may involve high-dimensional integrals that are analytically intractable. Gauss-Hermite quadrature (GHQ) approximation can be applied but is only suitable for low-dimensional random effects. Based on the Quasi-Monte Carlo (QMC) approximation, a heuristic approach is proposed to calculate the maximum likelihood estimates of parameters in the GLMM. The QMC points scattered uniformly on the high-dimensional integration domain are generated to replace the GHQ nodes. Compared to the GHQ approximation, the proposed method has many advantages such as its affordable computation, good approximation and fast convergence. Comparisons to the penalized quasi-likelihood estimation and Gibbs sampling are made using a real dataset and a simulation study. The real dataset is the salamander mating dataset whose modelling involves six 20-dimensional intractable integrals in the likelihood.     

Sustainable market valuation of buildings by the single-valued neutrosophic MAMVA method

Traditionally, the real estate asset assessment is performed by experienced valuators, who take into account its economic, social, physical and locational aspects. Nowadays, the construction industry is becoming more and more influenced by the sustainability requirements. Therefore, the inclusion of the sustainability evaluation into real estate asset valuation is of utmost importance. The Neutrosophic Multi-Attribute Market Value Assessment (MAMVA) method developed by the authors of this article handles market value calculations by solving multiple criteria assessment problems, and the initial information vagueness is modelled explicitly. The supplementary novelty of the present paper is the inclusion of the sustainability aspects into the real estate market valuation. The sustainable market valuation of Croydon University Hospital (Emergency Department) is performed as the case study to present numerical capabilities of the proposed approach. Our research findings suggest that neutrosophic MAMVA is a rational approach for calculations of property market valuation and might be suitable for application worldwide.     

The SIML estimation of realized volatility of the Nikkei-225 Futures and hedging coefficient with micro-market noise

For the estimation problem of the realized volatility and hedging coefficient by using high-frequency data with possibly micro-market noise, we use the Separating Information Maximum Likelihood (SIML) method, which was recently developed by Kunitomo and Sato [11], [12] and [13]. By analyzing the Nikkei-225 Futures data, we found that the estimates of realized volatility and the hedging coefficients have significant bias by using the traditional historical method which should be corrected. The SIML method can handle the bias problem in the estimation by removing the possible micro-market noise in multivariate high-frequency data. We show that the SIML method has the asymptotic robustness under non-Gaussian cases even when the market noises are autocorrelated and endogenous with the efficient market price or the signal term.     

Quasi-Monte Carlo estimation in generalized linear mixed models

Generalized linear mixed models (GLMMs) are useful for modelling longitudinal and clustered data, but parameter estimation is very challenging because the likelihood may involve high-dimensional integrals that are analytically intractable. Gauss–Hermite quadrature (GHQ) approximation can be applied but is only suitable for low-dimensional random effects. Based on the Quasi-Monte Carlo (QMC) approximation, a heuristic approach is proposed to calculate the maximum likelihood estimates of parameters in the GLMM. The QMC points scattered uniformly on the high-dimensional integration domain are generated to replace the GHQ nodes. Compared to the GHQ approximation, the proposed method has many advantages such as its affordable computation, good approximation and fast convergence. Comparisons to the penalized quasi-likelihood estimation and Gibbs sampling are made using a real dataset and a simulation study. The real dataset is the salamander mating dataset whose modelling involves six 20-dimensional intractable integrals in the likelihood.     

房地产经济投资决策仿真分析研究

该文以房地产市场需求众多影响因素，用线性多元回归数理统计作仿真分析研究，为房地产经济投资决策做量化参考，文章引用了一组实际调查的数据。     

Images as sets of locally weighted features

This paper presents a generic framework in which images are modelled as order-less sets of weighted visual features. Each visual feature is associated with a weight factor that may inform its relevance. This framework can be applied to various bag-of-features approaches such as the bag-of-visual-word or the Fisher kernel representations. We suggest that if dense sampling is used, different schemes to weight local features can be evaluated, leading to results that are often better than the combination of multiple sampling schemes, at a much lower computational cost, because the features are extracted only once. This allows our framework to be a test-bed for saliency estimation methods in image categorisation tasks. We explored two main possibilities for the estimation of local feature relevance. The first one is based on the use of saliency maps obtained from human feedback, either by gaze tracking or by mouse clicks. The method is able to profit from such maps, leading to a significant improvement in categorisation performance. The second possibility is based on automatic saliency estimation methods, including Itti & Koch’s method and SIFT’s DoG. We evaluated the proposed framework and saliency estimation methods using an in house dataset and the PASCAL VOC 2008/2007 dataset, showing that some of the saliency estimation methods lead to a significant performance improvement in comparison to the standard unweighted representation.     

融合时序特征约束与联合优化的 点云3维人体姿态序列估计

目的 3维人体姿态估计传统方法通常采用单帧点云作为输入,可能会忽略人体运动平滑度的固有先验知识,导致产生抖动伪影。目前,获取2维人体姿态标注的真实图像数据集相对容易,而采集大规模的具有高质量3维人体姿态标注的真实图像数据集进行完全监督训练有一定难度。对此,本文提出了一种新的点云序列3维人体姿态估计方法。方法 首先从深度图像序列估计姿态相关点云,然后利用时序信息构建神经网络,对姿态相关点云序列的时空特征进行编码。选用弱监督深度学习,以利用大量的更容易获得的带2维人体姿态标注的数据集。最后采用多任务网络对人体姿态估计和人体运动预测进行联合训练,提高优化效果。结果 在两个数据集上对本文算法进行评估。在ITOP(invariant-top view dataset)数据集上,本文方法的平均精度均值(mean average precision,mAP)比对比方法分别高0.99%、13.18%和17.96%。在NTU-RGBD数据集上,本文方法的mAP值比最先进的WSM(weakly supervised adversarial learning methods)方法高7.03%。同时,在ITOP数据集上对模型进行消融实验,验证了算法各个不同组成部分的有效性。与单任务模型训练相比,多任务网络联合进行人体姿态估计和运动预测的mAP可以提高2%以上。结论 本文提出的点云序列3维人体姿态估计方法能充分利用人体运动连续性的先验知识,获得更平滑的人体姿态估计结果,在ITOP和NTU-RGBD数据集上都能获得很好的效果。采用多任务网络联合优化策略,人体姿态估计和运动预测两个任务联合优化求解,有互相促进的作用。