Search results

Social signal processing under affective computing aims at recognizing and extracting useful human social interaction patterns. Fight is a common social interaction in real life. A fight detection system finds wide applications. This paper aims to detect fights in a natural and low-cost manner.

Research works on fight detection are often based on visual features, demanding substantive computation and good video quality. In this paper, the authors propose an approach to detect fight events through motion analysis. Most existing works evaluated their algorithms on public data sets manifesting simulated fights, where the fights are acted out by actors. To evaluate real fights, the authors collected videos involving real fights to form a data set. Based on the two types of data sets, the authors evaluated the performance of their motion signal analysis algorithm, which was then compared with the state-of-the-art approach based on MoSIFT descriptors with Bag-of-Words mechanism, and basic motion signal analysis with Bag-of-Words.

The experimental results indicate that the proposed approach accurately detects fights in real scenarios and performs better than the MoSIFT approach.

By collecting and annotating real surveillance videos containing real fight events and augmenting with well-known data sets, the authors proposed, implemented and evaluated a low computation approach, comparing it with the state-of-the-art approach. The authors uncovered some fundamental differences between real and simulated fights and initiated a new study in discriminating real against simulated fight events, with very good performance.

Up-to-date, the simulation of pedestrian behavior is used to support the design and analysis of urban infrastructure and public facilities. The purpose of this paper is to present a microscopic model that describes pedestrian behavior in a two-dimensional space. It is based on multi-agent systems and cellular automata theory. The concept of layered-intelligent terrain from the video game industry is reused and concepts such as tracing, evasion and rejection effects related to pedestrian interactive behavior are involved. In a simulation scenario, an agent represents a pedestrian with homogeneous physical characteristics such as walking speed and height. The agents are moved through a discrete space formed by a lattice of hexagonal cells, where each one can contain up to one agent at the same time. The model was validated by using a test that is composed of 17 real data sets of pedestrian unidirectional flow. Each data set has been extracted from laboratory-controlled scenarios carried out with up to 400 people walking through a corridor whose configuration changed in form of the amplitude of its entrance doors and the amplitude of its exit doors from one experiment to another. Moreover, each data set contained different groups of coordinates that compose pedestrian trajectories. The scenarios were replicated and simulated using the proposed model, obtaining 17 simulated data sets. In addition, a measurement methodology based on Voronoi diagrams was used to compute the velocity, density and specific flow of pedestrians to build a time-series graphic and a set of heat maps for each of the real and simulated data sets.

The approach consists of a multi-agent system and cellular automata theory. The obtained results were compared with other studies and a statistical analysis based on similarity measurement is presented.

A microscopic mobility model that describes pedestrian behavior in a two-dimensional space is presented. It is based on multi-agent systems and cellular automata theory. The concept of layered-intelligent terrain from the video game industry is reused and concepts such as tracing, evasion and rejection effects related to pedestrian interactive behavior are involved. On average, the simulated data sets are similar by 82 per cent in density and 62 per cent in velocity compared to the real data sets. It was observed that the relation between velocity and density from real scenarios could not be replicated.

The main limitations are presented in the speed simulations. Although the obtained results present a similar behavior to the reality, it is necessary to introduce more variables in the model to improve the precision and calibration. Other limitation is the dimension for simulating variables at this moment 2D is presented. So the resolution of cells, making that pedestrian to occupy many cells at the same time and the addition of three dimensions to the terrain will be a good challenge.

In total, 17 data sets were generated as a case study. They contain information related to speed, trajectories, initial and ending points. The data sets were used to calibrate the model and analyze the behavior of pedestrians. Geospatial data were used to simulate the public infrastructure in which pedestrians navigate, taking into account the initial and ending points.

The social impact is directly related to the behavior analysis of pedestrians to know tendencies, trajectories and other features that aid to improve the public facilities. The results could be used to generate policies oriented toward developing more consciousness in the public infrastructure development.

The general methodology is the main value of this work. Many approaches were used, designed and implemented for analyzing the pedestrians’ behavior. In addition, all the methods were implemented in plug-in for Quantum GIS. The analysis was described with heat maps and statistical approaches. In addition, the obtained results are focused on analyzing the density, speed and the relationship between these features.

The purpose of this paper is to present a new data set of aerial imagery from robotics simulator (AIR). AIR data set aims to provide a starting point for localization system development and to become a typical benchmark for accuracy comparison of map-based localization algorithms, visual odometry and SLAM for high-altitude flights.

The presented data set contains over 100,000 aerial images captured from Gazebo robotics simulator using orthophoto maps as a ground plane. Flights with three different trajectories are performed on maps from urban and forest environment at different altitudes, totaling over 33 kilometers of flight distance.

The review of previous research studies show that the presented data set is the largest currently available public data set with downward facing camera imagery.

This paper presents the problem of missing publicly available data sets for high-altitude (100‒3,000 meters) UAV flights; the current state-of-the-art research studies performed to develop map-based localization system for UAVs depend on real-life test flights and custom-simulated data sets for accuracy evaluation of the algorithms. The presented new data set solves this problem and aims to help the researchers to improve and benchmark new algorithms for high-altitude flights.

It has long been challenged that the distributions of empirical returns do not follow the log-normal distribution upon which many celebrated results of finance are based including the Black–Scholes Option-Pricing model. Borland (2002) succeeds in obtaining alternate closed form solutions for European options based on Tsallis distribution, which allow for statistical feedback as a model of the underlying stock returns. Motivated by this, we simulate two distinct time series based on initial data from NIFTY daily close values, one based on the Gaussian return distribution and the other on non-Gaussian distribution. Using techniques of non-linear dynamics, we examine the underlying dynamic characteristics of both the simulated time series and compare them with the characteristics of actual data. Our findings give a definite edge to the non-Gaussian model over the Gaussian one.

A new nonparametric procedure is developed to evaluate the significance of violations of weak separability. The procedure correctly detects weak separability with high probability using simulated data that have violations of weak separability caused by adding measurement error. Results are not very sensitive when the amount of measurement error is miss-specified by the researcher. The methodology also correctly rejects weak separability for nonseparable simulated data. We fail to reject weak separability for a monetary and consumption data set that has violations of revealed preference, which suggests that measurement error may be the source of the observed violations.

The objective of this research is to examine, validate, and recommend techniques for handling the problem of missingness in observational data. We use a rich observational data set, the Nielsen HomeScan data set, which allows us to effectively combine elements from simulated data sets: large numbers of observations, large number of data sets and variables, allowing elements of “design” that typically come with simulated data, and its observational nature. We created random 20% and 50% uniform missingness in our data sets and employed several widely used methods of single imputation, such as mean, regression, and stochastic regression imputations, and multiple imputation methods to fill in the data gaps. We compared these methods by measuring the error of predicting the missing values and the parameter estimates from the subsequent regression analysis using the imputed values. We also compared coverage or the percentages of intervals that covered the true parameter in both cases. Based on our results, the method of single regression or conditional mean imputation provided the best predictions of the missing price values with 28.34 and 28.59 mean absolute percent errors in 20% and 50% missingness settings, respectively. The imputation from conditional distribution method had the best rate of coverage. The parameter estimates based on data sets imputed by conditional mean method were consistently unbiased and had the smallest standard deviations. The multiple imputation methods had the best coverage of both the parameter estimates and predictions of the dependent variable.

This chapter compares the performance of the maximum simulated likelihood (MSL) approach with the composite marginal likelihood (CML) approach in multivariate ordered-response situations. The ability of the two approaches to recover model parameters in simulated data sets is examined, as is the efficiency of estimated parameters and computational cost. Overall, the simulation results demonstrate the ability of the CML approach to recover the parameters very well in a 5–6 dimensional ordered-response choice model context. In addition, the CML recovers parameters as well as the MSL estimation approach in the simulation contexts used in this study, while also doing so at a substantially reduced computational cost. Further, any reduction in the efficiency of the CML approach relative to the MSL approach is in the range of nonexistent to small. When taken together with its conceptual and implementation simplicity, the CML approach appears to be a promising approach for the estimation of not only the multivariate ordered-response model considered here, but also for other analytically intractable econometric models.

This paper aims to put forward and compare two accessible approaches to model and forecast spot prices in the fishing industry. The first modelling approach is a Markov-switching model (MSM) in which a Markov chain captures different economic regimes and a stochastic convenience yield is embedded in the spot price. The second approach is based on a multi-factor model (MFM) featuring three correlated stochastic factors.

The two proposed approaches are analysed in terms of parameter-estimation accuracy, information criteria and prediction performance. For MSM’s calibration, the quasi-log-likelihood method was applied directly while for the MFM’s parameter estimation, this paper designs an enhanced multi-variate maximum likelihood method with the aid of moments matching. The numerical experiments make use of both simulated and actual data compiled by the Fish Pool ASA. Data on both the Fish Pool’s forwards and Norwegian T-bill yields were additionally used in the MFM’s implementation.

Using simulated data sets, the MSM estimation gives more accurate results than the MFM estimation in terms of the norm in ℓ² between the “true” and “computed” parameter estimates and significantly lower standard errors. With actual data sets used to evaluate the forecast values, both approaches have similar performances based on the error analysis. Under some metrics balancing goodness of fit and model complexity, the MFM outperforms the MSM.

With the aid of simulated and observed data sets examined in this paper, insights are gained concerning the appropriateness, as well as the benefits and weaknesses of the two proposed approaches. The modelling and estimation methodologies serve as prelude to reliable frameworks that will support the pricing and risk management of derivative contracts on fish price evolution, which creates price risk transfer mechanisms from the fisheries/aquaculture sector to the financial industry.

The purpose of this paper is to investigate an efficient means of estimating the ability of students solving problems in the computer-based learning environment.

Item response theory (IRT) and TrueSkill were applied to simulated and real problem solving data to estimate the ability of students solving homework problems in the massive open online course (MOOC). Based on the estimated ability, data mining models predicting whether students can correctly solve homework and quiz problems in the MOOC were developed. The predictive power of IRT- and TrueSkill-based data mining models was compared in terms of Area Under the receiver operating characteristic Curve.

The correlation between students’ ability estimated from IRT and TrueSkill was strong. In addition, IRT- and TrueSkill-based data mining models showed a comparable predictive power when the data included a large number of students. While IRT failed to estimate students’ ability and could not predict their problem solving performance when the data included a small number of students, TrueSkill did not experience such problems.

Estimating students’ ability is critical to determine the most appropriate time for providing instructional scaffolding in the computer-based learning environment. The findings of this study suggest that TrueSkill can be an efficient means for estimating the ability of students solving problems in the computer-based learning environment regardless of the number of students.