An Innovated Integrated Model Using Singular Spectrum Analysis and Support Vector Regression Optimized by Intelligent Algorithm for Rainfall Forecasting

Rainfall forecasting is becoming more and more significant and precipitation anomalies would lead to droughts and floods disasters.However,because of the complexity and non-stationary of rainfall data,it is difficult to forecast.In this paper,a novel hybrid model to forecast rainfall is developed by incorporating singular spectrum analysis (SSA) and dragonfly algorithm (DA) into support vector regression (SVR) method.Firstly,SSA is used for extracting the trend components of the hydrological data.Then,SVR is utilized to deal with the volatility and irregularity of the precipitation series.Finally,the parameter of SVR is optimized by DA.The proposed SSA-DA-SVR method is used to forecast the monthly precipitation for Songbai,Panshui,Lanma and Jiulongchi stations.To validate the efficiency of the method,four compared models,DA-SVR,SSA-GWO-SVR,SSA-PSO-SVR and SSA-CS-SVR are established.The result shows that the proposed method has the best performance among all five models,and its prediction has high precision and accuracy.

Learning Hand Latent Features for Unsupervised 3D Hand Pose Estimation

Recent hand pose estimation methods require large numbers of annotated training data to extract the dynamic information from a hand representation.Nevertheless,precise and dense annotation on the real data is difficult to come by and the amount of information passed to the training algorithm is significantly higher.This paper presents an approach to developing a hand pose estimation system which can accurately regress a 3D pose in an unsupervised manner.The whole process is performed in three stages.Firstly,the hand is modelled by a novel latent tree dependency model (LTDM) which transforms internal joints location to an explicit representation.Secondly,we perform predictive coding of image sequences of hand poses in order to capture latent features underlying a given image without supervision.A mapping is then performed between an image depth and a generated representation.Thirdly,the hand joints are regressed using convolutional neural networks to finally estimate the latent pose given some depth map.Finally,an unsupervised error term which is a part of the recurrent architecture ensures smooth estimation of the final pose.To demonstrate the performance of the proposed system,a complete experiment was conducted on three challenging public datasets,ICVL,MSRA,and NYU.The empirical results show the significant performance of our method which is comparable or better than the state-of-the-art approaches.

Loader and Tester Swarming Drones for Cellular PhoneNetwork Loading and Field Test: Non-stochasticParticle Swarm Optimization

Cellular network operators have problems to test their network without affecting their user experience. Testingnetwork performance in a loaded situation is a challenge for the network operator because network performance differswhen it has more load on the radio access part. Therefore, in this paper, deploying swarming drones is proposed to loadthe cellular network and scan/test the network performance more realistically. Besides, manual swarming dronenavigation is not efficient enough to detect problematic regions. Hence, particle swarm optimization is proposed to bedeployed on swarming drone to find the regions where there are performance issues. Swarming drone communicationshelps to deploy the particle swarm optimization (PSO) method on them. Loading and testing swarm separation help tohave almost non-stochastic received signal level as an objective function. Moreover, there are some situations that morethan one network parameter should be used to find a problematic region in the cellular network. It is also proposed toapply multi-objective PSO to find more multi-parameter network optimization at the same time.

Flame Recognition in Video Images with Color and Dynamic Features of Flames

Recently,video based flame detection has become an important approach for early detection of fire under complex circumstances.However,the detection accuracy of most existing methods remains unsatisfactory.In this paper,we develop a new algorithm that can significantly improve the accuracy of flame detection in video images.The algorithm segments a video image and obtains areas that may contain flames by combining a two-step clustering based approach with the RGB color model.A few new dynamic and hierarchical features associated with the suspected regions,including the flicker frequency of flames,are then extracted and analyzed.The algorithm determines whether a suspected region contains flames or not by processing the color and dynamic features of the area altogether with a classifier,which can be a BP neural network,a k nearest neighbor classifier or a support vector machine.Testing results show that this algorithm is robust and efficient,and is able to significantly reduce the probability of false alarms.

Intelligent fish tank based on WiFi module

In the paper,the intelligent fish tank using STC89C52 as the control core embedded HC-SR04 ultrasonic distance measurement module and DS18B20 temperature sensor is introduced.This system can be used to remotely control and collect the data of the temperature and the level of water in the fish tank through WiFi module (ESP8266-01).When the water level is less than the default value,the system will be adjusted by adding water into the tank.At the same time,people could also get the data and control the tank whenever they want.The micro-controller is connected to the Internet through the WiFi module.With the help of MicroPython firmware,python programs are compiled within this WiFi module in order to connect to the WiFi at home,providing data transfer function.Android smart phones could connect to this system through WiFi and send commands.In this way,the fish tank could be controlled remotely to ensure the stability of the water temperature and level in the tank.

Toward Global Complex Systems Control - The Autonomous Intelligence Challenge

Complex systems are the emerging new scientific frontier with modern technology advance and new parametric domains study in natural systems.An important challenge is,contrary to classical systems studied so far,the great difficulty in predicting their future behaviour from initial time because,by their very structure,interactions strength between system components is shielding completely their specific individual features.Independent of clear existence of strict laws complex systems are obeying like classical systems,it is however possible today to develop methods allowing to handle dynamical properties of such systems and to master their evolution.So the methods should be imperatively adapted to representing system self organization when becoming complex.This rests upon the new paradigm of passing from classical trajectory space to more abstract trajectory manifolds associated to natural system invariants characterizing complex system dynamics.The methods are basically of qualitative nature,independent of system state space dimension and,because of its generic impreciseness,privileging robustness to compensate for not well known system parameters and functional variations.This points toward the importance of control approach for complex system study in adequate function spaces,the more as for industrial applications there is now evidence that transforming a complicated man made system into a complex one is extremely beneficial for overall performance improvement.But this last step requires larger intelligence delegation to the system requiring more autonomy for exploiting its full potential.A well-defined,meaningful and explicit control law should be set by using equivalence classes within which system dynamics are forced to stay,so that a complex system described in very general terms can behave in a prescribed way for fixed system parameters value.Along the line traced by Nature for living creatures,the delegation is expressed at lower level by a change from regular trajectory space control to task space control following system reassessment into its complex stage imposed by the high level of interactions between system constitutive components.Aspects of this situation with coordinated action on both power and information fluxes are handled in a new and explicit control structure derived from application of Fixed Point Theorem which turns out to better perform than (also explicit) extension of Popov criterion to more general nonlinear monotonically upper bounded potentials bounding system dynamics discussed here.An interesting observation is that when correctly amended as proposed here,complex systems are not as commonly believed a counterexample to reductionism so strongly influential in Science with Cartesian method supposedly only valid for complicated systems.

A Perspective of Conventional and Bio-inspired Optimization Techniques in Maximum Likelihood Parameter Estimation

Maximum likelihood estimation is a method of estimating the parameters of a statistical model in statistics. It has been widely used in a good many multi-disciplines such as econometrics, data modelling in nuclear and particle physics, and geographical satellite image classification, and so forth. Over the past decade, although many conventional numerical approximation approaches have been most successfully developed to solve the problems of maximum likelihood parameter estimation, bio-inspired optimization techniques have shown promising performance and gained an incredible recognition as an attractive solution to such problems. This review paper attempts to offer a comprehensive perspective of conventional and bio-inspired optimization techniques in maximum likelihood parameter estimation so as to highlight the challenges and key issues and encourage the researches for further progress.

Reinforcement Learning:A Technical Introduction–Part I

Reinforcement learning provides a cognitive science perspective to behavior and sequential decision making providedthat reinforcement learning algorithms introduce a computational concept of agency to the learning problem.Hence it addresses an abstract class of problems that can be characterized as follows: An algorithm confronted withinformation from an unknown environment is supposed to find step wise an optimal way to behave based only on somesparse, delayed or noisy feedback from some environment, that changes according to the algorithm’s behavior. Hencereinforcement learning offers an abstraction to the problem of goal-directed learning from interaction. The paper offersan opinionated introduction in the algorithmic advantages and drawbacks of several algorithmic approaches to providealgorithmic design options.

Recent Advances in Particle Swarm Optimization for Large Scale Problems

Accompanied by the advent of current big data ages,the scales of real world optimization problems with many decisive design variables are becoming much larger.Up to date,how to develop new optimization algorithms for these large scale problems and how to expand the scalability of existing optimization algorithms have posed further challenges in the domain of bio-inspired computation.So addressing these complex large scale problems to produce truly useful results is one of the presently hottest topics.As a branch of the swarm intelligence based algorithms,particle swarm optimization (PSO) for coping with large scale problems and its expansively diverse applications have been in rapid development over the last decade years.This reviewpaper mainly presents its recent achievements and trends,and also highlights the existing unsolved challenging problems and key issues with a huge impact in order to encourage further more research in both large scale PSO theories and their applications in the forthcoming years.

Classification of the Priority of Auditing XBRL Instance Documents with Fuzzy Support Vector Machines Algorithm

Concluding the conformity of XBRL(eXtensible Business Reporting Language)instance documents law to the Benford's law yields different results before and after a company's financial distress.A new idea of applying the machine learning technique to redefine the way conventional auditors work is therefore proposed since the unacceptable conformity implies a large likelihood of a fraudulent document.Fuzzy support vector machines models are developed to implement such an idea.The dependent variable is a fuzzy variable quantifying the conformity of an XBRL instance document to the Benford's law;whereas,independent variables are financial ratios.The interval factor method is introduced to express the fuzziness in input data.It is found the range of a fuzzy support vector machines model is controlled by maximum and minimum dependent and independent variables.Therefore,defining any member function to describe the fuzziness in input data is unnecessary.The results of this study indicate that the price-to-book ratio versus equity ratio is suitable to classify the priority of auditing XBRL instance documents with the less than 30%misclassification rate.In conclusion,the machine learning technique may be used to redefine the way conventional auditors work.This study provides the main evidence of applying a future project of training smart auditors.

PC-VINS-Mono: A Robust Mono Visual-Inertial Odometry with Photometric Calibration

Feature detection and Tracking, which heavily rely on the gray value information of images, is a very importance procedure for Visual-Inertial Odometry (VIO) and the tracking results significantly affect the accuracy of the estimation results and the robustness of VIO. In high contrast lighting condition environment, images captured by auto exposure camera shows frequently change with its exposure time. As a result, the gray value of the same feature in the image show vary from frame to frame, which poses large challenge to the feature detection and tracking procedure. Moreover, this problem further been aggravated by the nonlinear camera response function and lens attenuation. However, very few VIO methods take full advantage of photometric camera calibration and discuss the influence of photometric calibration to the VIO. In this paper, we proposed a robust monocular visual-inertial odometry, PC-VINS-Mono, which can be understood as an extension of the opens-source VIO pipeline, VINS-Mono, with the capability of photometric calibration. We evaluate the proposed algorithm with the public dataset. Experimental results show that, with photometric calibration, our algorithm achieves better performance comparing to the VINS-Mono.

Security Challenges and Countermeasures for the Heterogeneity of IoT Applications

The Internet of Things (IoT) enables the integration of data from virtual and physical worlds. It involves smart objects that can understand and react to their environment in a variety of industrial, commercial and household settings. As the IoT expands the number of connected devices, there is the potential to allow cyber-attackers into the physical world in which we live, as they seize on security holes in these new systems. New security issues arise through the heterogeneity of IoT applications and devices and their large-scale deployment.

A comprehensive analysis of smart home energy management system optimization techniques

Development of smart grid technology provides an opportunity to various consumers in context for scheduling their energy utilization pattern by themselves.The main aim of this whole exercise is to minimize energy utilization and reduce the peak to average ratio (PAR) of power.The two way flow of information between electric utilities and consumers in smart grid opened new areas of applications.The main component is this management system is energy management controller (EMC),which collects demand response (DR) i.e.real time energy price from various appliances through the home gateway (HG).An optimum energy scheduling pattern is achieved by EMC through the utilization of DR information.This optimum energy schedule is provided to various appliances via HG.The rooftop photovoltaic system used as local generation micro grid in the home and can be integrated to the national grid.Under such energy management scheme,whenever solar generation is more than the home appliances energy demand,extra power is supplied back to the grid.Consequently,different appliances in consumer premises run in the most efficient way in terms of money.Therefore this work provides the comprehensive review of different smart home appliances optimization techniques,which are based on mathematical and heuristic one.

An Experimental Analysis of the Applications of Datamining Methods on Bigdata

Data mining is a procedure of separating covered up,obscure,however possibly valuable data from gigantic data.Huge Data impactsly affects logical disclosures and worth creation.Data mining(DM)with Big Data has been broadly utilized in the lifecycle of electronic items that range from the structure and generation stages to the administration organize.A far reaching examination of DM with Big Data and a survey of its application in the phases of its lifecycle won't just profit scientists to create solid research.As of late huge data have turned into a trendy expression,which constrained the analysts to extend the current data mining methods to adapt to the advanced idea of data and to grow new scientific procedures.In this paper,we build up an exact assessment technique dependent on the standard of Design of Experiment.We apply this technique to assess data mining instruments and AI calculations towards structure huge data examination for media transmission checking data.Two contextual investigations are directed to give bits of knowledge of relations between the necessities of data examination and the decision of an instrument or calculation with regards to data investigation work processes.

Shaping the Next Generation Pharmaceutical Supply Chain Control Tower with Autonomous Intelligence

This paper summarizes the findings of an industry panel study evaluating how new Autonomous Intelligence technologies,such as artificial intelligence and machine learning,impact the system and operational architecture of supply chain control tower (CT) implementations that serve the pharmaceutical industry.Such technologies can shift CTs to a model in which real-time information gathering,analysis,and decision making are possible.This can be achieved by leveraging these technologies to better manage decision complexity and execute decisions at levels that cannot otherwise be managed easily by humans.Some of the key points identified are in the areas of the fundamental capabilities that need to be supported and the improved level of decision visibility that they provide.We also consider some the challenges in achieving this,which include data quality and integrity,collaboration and data sharing across supply chain tiers,cross-system interoperability,decision-validation and organizational impacts,among others.

The autonomous intelligence challenge

The considerable development of modern technology during last decades has been accompanied by improvements in associated adapted machines’ performance.The improvement process has always been guided by the same rules of researching higher efficiency and more secure effects each time.This leads to a progressive transfer of human action to more adapted and more specific effecting objects,from simple tools for elementary actions to more sophisticated machines for quite complex tasks.Each step of this transfer of human operator action has been realized by delegating to the effecting machine efficiency,accuracy,power and safety—basically all of technical nature and linked to power flux,with the human operator still keeping the mastery of the action to fulfil his own goals.

Monitoring on Guardrails to Afford Road Safety Using IOT

The preservation of the road infrastructures has become an important issue to the road safety and structural monitoring systems industry aiming to reduce the maintenance cost and also to increase the drivers safety. The collision features of the simulation of car- guard rail. It is found that the vibration features of the guardrail within the accident have a good performance to the accident identification. The vibration data of the guardrail are recorded real-timely by the nodes with accelerator sensors on the guardrail network. Then the collision accident is identified in terms of the vibration threshold. The proposed design is a system which can detect accidents in significantly less time and sends the basic information to first aid center within a few seconds covering geographical coordinate the time and angle in which a vehicle accident had occurred. This alert message is sent to the rescue team in a short time which will help in saving the valuable lives.

Outdoor Temperature Estimation Using ANFIS for Soft Sensors

In recent years,several studies using smart methods and soft computing in the field of HVAC systems have been provided.In this paper,we propose a framework which will strengthen the benefits of the Fuzzy Logic(FL)and Neural Fuzzy(NF)systems to estimate outdoor temperature.In this regard,Adaptive Neuro Fuzzy Inference System(ANFIS)is used in effective combination of strategic information for estimating the outdoor temperature of the building.A novel versatile calculation focused around ANFIS is proposed to adjust logical progressions and to weaken the questionable aggravation of estimation information from multisensory.Due to ANFIS accuracy in specialized predictions,it is an effective device to manage vulnerabilities of each experiential framework.The NF system can concentrate on measurable properties of the samples throughout the preparation sessions.Reproduction results demonstrate that the calculation can successfully alter the framework to adjust context oriented progressions and has solid combination capacity in opposing questionable data.This sagacious estimator is actualized utilizing Matlab and the exhibitions are explored.The aim of this study is to improve the overall performance of HVAC systems in terms of energy efficiency and thermal comfort in the building.

New Combined Method to Improve Arabic POS Tagging

One of the important tasks in Natural language processing is the part of speech tagging. For the Arabic language we have a lot of works but their performances do not rise to the required level, due to the complexity of the task and the Arabic language characteristics. In this work we study a combination between two different approaches for Arabic POSTaggers. The first one is a maximum entropy-based one, and the second is a statistical/rule-based one. Fur-thermore, we add a knowledge-based method to annotate Arabic particles. Our idea improves the accuracy rate. We passed from almost 85% to almost 90% using our combined method, which seem promoter.

Neural processor in artificial intelligence advancement

A neuron network is a computational model based on structure and functions of biological neural networks.Information that flows through the network affects the structure of the neuron network because neural network changesor learns,in a sense-based on that input and output.Although neural network being highly complex (for example change of weights for every new data within the time frame) an experimental model of high level architecture of neural processor is proposed.Neural Processor performs all the functions that an ordinary neural network does like adaptive learning,self-organization,real time operations and fault tolerance.In this paper,analysis of neural processing is discussed and presented with experiments,graphical representation including data analysis.