An Application of Neural Network on Traffic Speed Prediction Under Adverse Weather Condition

Size: px
Start display at page:

Download "An Application of Neural Network on Traffic Speed Prediction Under Adverse Weather Condition"

Transcription

1 See discussions, stats, and author profiles for this publication at: An Application of Neural Network on Traffic Speed Prediction Under Adverse Weather Condition Article June 1995 CITATIONS 37 READS authors: Shan-Huen Huang National Kaohsiung First University of Science and Technology 15 PUBLICATIONS 256 CITATIONS SEE PROFILE Bin Ran University of Wisconsin Madison 287 PUBLICATIONS 2,387 CITATIONS SEE PROFILE Some of the authors of this publication are also working on these related projects: 2012CB View project V-SPOC Traffic Data Portal View project All content following this page was uploaded by Bin Ran on 23 December 2014 The user has requested enhancement of the downloaded file

2 An Application of Neural Network on Traffic Speed Prediction Under Adverse Weather Condition S H Huang and Bin Ran Abstract A neural network model for predicting the traffic speed under adverse weather conditions is proposed One link located in Chicago was chosen and all the data involved was collected from the Internet The Back Propagation algorithm was used to train the neural network model for approaching the best prediction results The MATLAB software was used to solve this model The results has demonstrated that, neural network is an effective tool theory to predict traffic situation if appropriate model architecture and input data are available 1 Introduction and Objectives The impact of adverse weather on traffic speed can be severe so as to influence motorists driving significantly On the other hand, traffic speed prediction is an important tool to support Intelligent Transportation System (ITS) With precise traffic speed prediction, travelers can easily obtain the travel time information so as to arrange their schedule by using a traveler information system Moreover, traffic managers can exploit the prediction information to deploy various traffic management strategies as well A lot of factors could affect the traffic speed including Construction, Incident and Weather Each one of them will reduce the vehicle speed by different level s severity The reduction of speed caused by those factors can be termed impact How to procure the traffic impact or predict the traffic speed under adverse weather conditions is the main objective of this paper In this paper, a Neural Network based on back- propagation algorithm is proposed for predicting the impact of adverse weather 1

3 MATLAB, a powerful Mathematics software, is adopted to apply the model All the raw data are acquired from the Internet 2 Literature Review Keeping traffic moving is the biggest challenge that all levels of government are facing worldwide Nowadays, Advanced Traveler Information System (ATIS) is becoming more important to allow road users avoiding congestion Several ATIS systems like The TravTek System and The Navmate System were developed to provide road users the updated information and guide them selecting the shortest or fastest routes ( Traffic situation prediction is always a challenge in transportation field Several different methods are used to predict the traffic situation and avoid the congestion Traffic Density Prediction Map (TDPM) is one of the prediction methods, which is an interactive map with a time slider It indicates the predicted traffic delay at the time indicated by the slider Traffic density is predicted from several variables such as season and day of week, combined with historic traffic data from road sensors ( Other research methods have been proposed as well using other theories such as statistics, fuzzy, neural network, etc Texas A&M University investigated the deterministic properties of traffic flow using a nonlinear time series analysis technique The study concluded that the traffic data exhibits chaotic properties and techniques based on phase space dynamics can be used to analyze and predict the traffic flow Helsinki University developed the models to predict the speed and flow 15 minutes ahead of the observation period in five-minute periods (Nair) The purpose of this research was to study the influence of various factors on the results of the short-time prediction of the traffic situation on motorways Multi-layer perceptron networks were used as prediction modes (Innamaa, 2000) 2

4 On the other hand, the Road Weather Information System (RWIS) technology has been proven to improve winter travel safety while reducing both the amount of deicing chemical and the cost of maintaining safer roads and runways for the traveling public ( Many state DOTs have tried to provide the road weather information for road users to improve the safety and adjust the route ( ( 3 Traffic Speed Prediction Under Adverse Weather The definition of adverse weather here is the weather condition that will decrease the visibility and worsen the pavement condition Road users will slow down the vehicle in order to keep safe driving while the visibility is decreased or the pavement condition is bad Generally, the types of adverse weather could be classified as below: 1 Fog (Reduce Visibility) 2 Rain (Reduce Visibility and Worsen Pavement Condition) 3 Snow (Reduce Visibility and Worsen Pavement Condition) 4 Flooding (Worsen Pavement Condition) 5 Ice (Worsen Pavement Condition) Speed A B Link Capacity Fig 1 Speed-Flow Diagram Flow 3

5 Fig 1 shows a common speed-flow diagram and the solid line is a normal speed-flow relationship Speed decreases along with the increase of traffic flow till approaching the link capacity Then, speed keeps decreasing along with the decreasing traffic flow The dashed line indicates the speed-flow relationship while the link encounters the adverse weather condition As shown in the figure, the impact of the adverse weather sould be the length of segment AB In addition, the capacity of the link would be decreased because of the adverse weather 31 Assumption of the Prediction Procedure In order to predict the speed, there are some presumptions before it kicks off 1 The weather forecast is assumed exactly correct Because weather information is one of the parameters in prediction, it has to be valid Otherwise the prediction will have large errors 2 Part of the collected data that had been influenced by other factors (like construction, incident, etc) should be found and separated from this prediction Those data constitute another set of problems; only weather factor and normal traffic flow are discussed and concerned in this research 3 Traffic flow in a specific hour of one week is similar to another week s traffic flow in the same hour For example, the traffic flow at 7:00~8:00 on Monday is similar to the traffic flow at 7:00~8:00 of last Monday This assumption is not suited for long-term prediction because it doesn t make sense the traffic flow is similar to the traffic flow 1 or 2 years ago or later even though one has exactly the same day and the same hour That is because the city is changing, and the population and the life style keep changing over time Therefore, this model is only appropriate for short-term prediction Generally, the prediction time range shouldn t be over 1 week 4

6 32 The Scenarios The prediction models were developed for one link in the Chicago metropolitan area The weather and traffic flow information about this link have been collected for six months During this period, the weather conditions happened include snow, rain, fog, etc The weather data were gathered each hour and speed data were gathered every five minutes The link speed during each hour is the average speed of all the collected speeds during that hour The traffic situation in the Chicago Metropolitan area is quite complex However, since this research is focus on the influence of traffic speed by adverse weather, other situation won t be discussed in the study 33 Data Collection and Variables 331 Data Collection Historic traffic data is fundamental for prediction systems In this study, all the information could be retrieved from the Internet The data needed in this model could be roughly categorized into two types, speed and the weather condition at the same time Both of them could be downloaded from the Internet Several JAVA programs were developed to connect to the Internet and retrieve information every five minutes for speed data and every one-hour for weather information The information of weather condition consists of visibility, temperature, moisture, linguistic weather terms, etc There are quite a number of sensors were buried under the pavement by DOT (Department of Transportation) in order to count the passed vehicles and obtain the real time traffic speed Therefore, traffic speed data can be acquired from the official DOT s website The weather condition information could be acquired from the National Oceanic and Atmospheric Administration (NOAA) website That information obtained from the Internet will be used directly or parsed to be the desired format as input variables 5

7 Fig 2 The Internet resource of Weather Condition 6

8 Fig 3 The Internet Resource of Traffic Information Figure 2 and Figure 3 illustrate the Internet resource of weather and traffic information, respectively( 332 Variables According to Figure 1, the impacts are due to the interaction of weather and traffic flow Therefore, all the variables needed in this model could be categorized into two types, weather and traffic flow Both of them will be discussed below 7

9 3321 Traffic Flow According to the speed-flow relationship, traffic flow is definitely one of the important variables in this model However, it is very difficult to use it directly in the model That is because each traffic flow might be corresponding to two different traffic speeds If the model receives only a value of traffic flow, say 500 vph, it is impossible to decide whether it is congested or not Therefore, the time variable is used to represent traffic flow Although the number of variables is increased, it is much clearer to descript the traffic 3322 Weather Condition There are two main parameters to describe how the weather affects the traffic speed, which are visibility and pavement condition The value of visibility could be obtained from the Internet directly Therefore, visibility would definitely be one of the variables in this model It is not so convenient to define pavement condition That is because there is no well-established index to describe how the pavement condition is affected by the weather Therefore, other variables are used to represent the pavement condition Here, temperature and moisture are recruited The possible pavement conditions caused by adverse weather would be flooding, ice and snow on the pavement The moisture should be high when it is flooding and the temperature should be below 32 F when snow or ice covers the pavement That would be perplexing when the fog comes or a pretty cold day (below 32 F) arrives without snow on the pavement The temperature and moisture might be the same as it rains or snows The solution is to add one more indicating variable in the model For example, if it was raining, we then add one variable rain and assign the value 1 on it In this case, the model will recognize that it is rain and there is water on the pavement 8

10 To sum up, according to the information obtained from the Internet and discussed above, we have identified several dozens of input parameters and they are illustrated as follows Input 1 Visibility (miles) Input 2 Temperature (ºF) Input 3 Moisture (%) Input 4 ~ 13 Weather Conditions (including Rain, Light Rain, Snow, Light Snow, Fog, Mist and so on Each of them will have the value 1 if the situation happened, otherwise, the value will be assigned as 0 as defalt) Input 14~44 Time variables They are tow sequences of time variables, which are Date part and Time part The Date part is a sequence as Monday, Tuesday, Wednesday, Thursday, Friday, Saturday, Sunday and the Time part is as 0, 1, 2, 3,, 23 Another four variable is included for indicating the order of the quarter hour such as first 15 minutes or second 15 minutes These four variables and other time variables are having value either 0 or 1 as well to indicate the data time and imply the traffic flow Input 45 Real time speed (MPH) This information was obtained directly from the Internet and will be use to adjust the weight of all link in the network 4 Methodologies 41 Neural Network Model In this study, multi-layer perceptron (MLP) networks were used as prediction models because of their good results in previous studies (Smith and Demetsky, 1994, 1997 and S Lee, 1998) Neural Network is a powerful model in solving complex problems Since the neural network has natural potential of solving nonlinear problem and can easily achieve the input-output mapping, it is perfect to use it for solving the predicting problem (Haykin, 1999) Figure 4 is the architectural graph of a multilayer perceptron The layer consists of several nodes in the most left side called input layer Each node in the input layer 9

11 receives one input signal Those signals flow from left to right layer by layer The layer in the right hand side with only one neuron is called output layer The number of input neurons was equal to the number of input parameters and the number of neurons in the output layer was equal to the number of output parameters There are two different kinds of signals identified in this network: 1 Function Signals A function signal is an input signal that comes in at the input end of the network, propagates forward (neuron by neuron) through the network, and emerges at the output end of the network as an output signal The function signal is also referred to as the input signal 2 Error Signals An error signal originates at an output neuron of the network, and propagates backward (layer by layer) through the network We refer to it as an error signal because its computation by every neuron of the network involves an error-dependent function in one form or another These two signal flows of all data go forward and backward once called one epoch In every epoch, the model will adjust the neuron s weight based on minimizing the error flow After a lot of iterations, the error equation will become converged That means it is reaching the minima of the error surface In this sense, the output is closest to the expected response In this model, that means the predicted speed is closer to the real world speed as much as possible 10

12 Input signal Visibility Temperatur Moisture Weather Condition Variables: Rain, Snow, etc Output Layer Time Variables: Monday, 2:00~3:00, etc Output Signal First Last Input Hidden Hidden Layer Layer Layer Fig 4 A comprehensive multilayer perceptron architecture 11

13 42 Back-Propagation Algorithm Back-Propagation is one of the most powerful algorithms to train the multiplayer perceptron and obtain the weight of each link Basically, error back-propagation learning consists of two passes through different layers of the network: a forward pass and a backward pass In the forward pass, an activity pattern (input vector) is applied to the sensory nodes of the network, and its effect propagates through the network layer by layer Finally, a set of outputs is produced as the actual response of the network During the forward pass, the synaptic weights of the networks are all fixed During the backward pass, on the other hand, the synaptic weights are all adjusted in accordance with an errorcorrection rule Specifically, the actual response of the network is subtracted from a desired (target) response to produce an error signal This error signal is then propagated backward through the network, against the direction of synaptic connections-hence the name error back-propagation The synaptic weights are adjusted to make the actual response of the network move closer to the desired response in a statistical sense (Li) T ε ( n) ( d( n) y( n) ) ( d( n) y( n) ) () 1 = 2 1 N 2 ε = ε ( n) ( 2 ) T n= 1 43 Training Criteria And Model Settings In order to obtain the training results quickly and appropriately, there are some training criteria and model settings to be established They are discussed as follows: 1 Learning rate (η) The larger the learning rate η, the larger the weight changes in each epoch, and the quicker the network learns However, the size of the learning rate can also influence whether the network achieves a stable solution If we make the learning-rate parameter η too large in order to speed up the rate of learning, the resulting large changes in the synaptic weights assume such a form that the network may become unstable (ie, oscillatory) In this model, learning rate (η) 12

14 has been set as 01 ( mentum) 2 Momentum (α) A simple method of increasing the rate of learning yet avoiding the danger of instability has been proposed as a slight modification of the back propagation algorithm so that it includes a momentum term Applied to back propagation, the concept of momentum is that previous changes in the weights should influence the current direction of movement in weight space This concept is implemented by the revised weight-update rule: ( n) = α w ( n ) + ηδ ( n) y ( n) w 1 ji ji j i where α is usually a positive number called the momentum constant and δ j is the local gradient for neuron j Plus, w ji ( n) is the weight between neuron j and i at iteration n In this model, α is set as 08 3 Activation Function (ϕ()) The computation of the local gradient for each neuron of the multiplayer perceptron requires knowledge of the derivative of the activation function ϕ() For this derivative to exist, we require the function ϕ() to be continuous In basic terms, differentiability is the only requirement that an activation function has to satisfy In this model, Logistic Function has been chosen to be the activation function for each neuron It follows that ( v ( n) ) 1 ϕ j j = > 0 1+ exp ( av ( n) ) induced local field of neuron j j a and < v j ( n) <, where ( n) v j is the 4 Epochs: In this study, epoch was set as 1000 for each training 5 It is useful to preprocess the data before introducing it to the network In this model, the data were scaled to vary between [-1,1] 13

15 6 Training will be stopped if the mean squared error is less than 10e-5 7 The order of input data had been randomized before introducing it to the network 44 Analysis of Time Period Before we determine the final input feature and model structure, the time period should be analyzed In transportation, usually, the predicting time period is better to be shorter That means if we can predict the traffic situation every minute is ideal However, it is hardly to predict the traffic situation in such a short time period because the traffic condition is changing rapidly Hence, the time period need to be increased in length to retain the stability On the other hand, it is meaningless to predict the traffic situation in a long time period like tow hours even though the prediction is very precise That is because this predicting result is not practicable in traffic management and other related fields Therefore, to figure out the balance between stability and practicability become an important issue before we finalize the whole model According to the assumptions mentioned above, it is believed that this model is not able to predict the traffic speed in five-minute basis That is because there is no guarantee the traffic flow is remaining similar to the same time period before or later According the result of experiments, ten-minute period model was not able to produce reasonable and acceptable output Ultimately, the fifteen-minute model illustrated acceptable performance and was selected to be the main model type 45 Time-Series Model: The time series model is employed to reproduce the other traffic speed prediction model The model was developed to predict the observed speed by using fifteen minutes ahead of the observation period in five-minute periods That means there are three input factors that are the speed five minutes, ten minutes and fifteen minutes prior to the predicting period The establishment of this time-series model is similar to the model established 14

16 previously, using neuron network along with back propagation algorithm All the data preprocessing procedures are remaining the same as previous model Since the previous model is a fifteen-minute basis model and this time-series model is a five-minute basis model, fifteen-minute average speed should be attained by the time-series model in order to be corresponding to the fifteen-minute model The number of the neuron in the input layer is the same with the input feature and the number of neuron in the output layer is the same with the output feature This time-series model has been demonstrated as a short-term traffic situation prediction with precise output result The purpose of establishing this time-series model here is to inspect the output of previous model If previous model can produce even or better performance than the time-series model, then the output result is acceptable However, the time-serious model is limited on short-term prediction unlike the NN model that can predict both short-term and long-term Therefore, we can take more advantages from the NN predicting model Denote the data at instant k as y(k), where y may be a vector, then the above treatment can be described as y = NN( y ) + or yk+ = NN( yk, yk 1,, yk i ) k 1 k 1 K respectively, where NN() stands for the neural network forecaster and i is the number of successive observations This treatment considers the time series as a nonlinear time series and tends to generate a nonlinear "auto-regression" model to fit the series Figure 5 illustrated the comprehensive time series model architecture (Hu, 2001) y(t) y^ () t - e(t) MLP U(t) U(t-1) U(t-m) Fig 5 A Comprehensive Time Series Multilayer Perceptron Architecture 15

17 5 Case Study Output In total, 4214 eligible data were collected for this link and they are divided into 4189 training data points and 25 testing data points, respectively After 1000 epochs, MATLAB will show the error between real speed and predicted speed Figure 6 shows the relationship between the real speed and the estimated speed Each point means the average speed of one specific hour The difference between real speed and predicted speed ( S S p e ) is the error, where S r is the real speed and r = S p is the predicted speed In this set of data, the mean of the error is and standard deviation is The results generated by the time-series model were also shown on the graph The error mean is equal to and the error standard deviation is equal to Both models are giving the close performance That means the output result from the model is reasonable and acceptable 16

18 Fig 6 The output chart of real speed and estimated speed 17

19 Fig7 The mean square error vs epoch Figure 7 shows the mean square error versus epoch The upper line indicates the mean square error and the lower line indicates the goal which is 1e-005 It shows obviously the error converged very early and the mean square error was That means backpropagation algorithm is capable to reach the minimum of error surface effectively The results generated by the time-series model were also shown on the graph The error mean is equal to and the error standard deviation is equal to Both models are giving the close performance That means the output result from the NN model is reasonable and acceptable 6 Conclusions and Future Research This output demonstrated that Neural Network Model could help on predicting traffic speed in a little bit longer term than the Time Series Model with precise output Having the predicting traffic speed data, road users can easily adjust their route and plan their trip 18

20 and traffic managers can deploy and adapt their operations based on the prediction traffic situation This study shows that if we choose four hidden layers, we will obtain the closest results (mean error is the smallest) That may be because there were too many input features so that more hidden layers could train the network better Few predicting points on Figure 5 are not really close to the expected output That is because of insufficient historic data Since this model deals with an input-output mapping problem, sufficient historic traffic data are fundamental for prediction systems That means if we want to obtain more precise predicted traffic speed, historic data should be collected as much as possible It is promising to obtain much better output if there is more complete information for pavement condition or other weather situations This predicting procedure is not only suitable for predicting the traffic speed under adverse weather but also suitable for predicting the traffic speed while other conditions are involved, including construction and incident It is reasonable to substitute the variables of rain or snow for construction time; number of lanes closed and closed lanes distance Once all the influential variables are decided, the same procedure will be taken to obtain the predicted speeds It is possible to predict the traffic speed of a route or a network by expanding this model No matter how big the network is or how long the route is, they are all made of links If all the links historic information is available, then the traffic speed is predictable 7 Reference [1] Smith B and Demetsky M (1994) Short-Term Traffic Flow prediction: Neural Network Approach Transportation Research Record 1453 Pp

21 [2] Demetsky M and Smith B (1997) Traffic Flow Forecasting: Comparison of Modeling Approaches Journal of Transportation Engineering Vol 123, No 4, July / August 1997 Pp [3] Haykin S (1999) Neural Networks, A Comprehensive Foundation International Edition, Second Edition Prentice Hall International, Inc, USA 842 s [4] Innamaa S (2000) SHORT-TERM PREDICTION OF TRAFFIC SITUATION USING MLP-NEURAL NETWORKS ( [5] Lee S, Kim D, Kim J and Cho B (1998) Comparison of Models for predicting Shout-Term Travel Speeds Conference CD-ROM, 5 th World Congress on Intelligent Transport Systems, October 1998, Seoul, Korea 9 p ( f) [6] Li S H, Comparative Analysis of Backpropagation and Extended Kalman Filter in Pattern and Batch Forms for Training Neural Networks ( f) [7] Nair A S, Liu JC, Rilett L and Gupta S, Non-Linear Analysis of Traffic Flow, [8] [9] [10] 20

22 [11] [12] [13] 21 View publication stats

4. Multilayer Perceptrons

4. Multilayer Perceptrons 4. Multilayer Perceptrons This is a supervised error-correction learning algorithm. 1 4.1 Introduction A multilayer feedforward network consists of an input layer, one or more hidden layers, and an output

More information

Short Term Load Forecasting Based Artificial Neural Network

Short Term Load Forecasting Based Artificial Neural Network Short Term Load Forecasting Based Artificial Neural Network Dr. Adel M. Dakhil Department of Electrical Engineering Misan University Iraq- Misan Dr.adelmanaa@gmail.com Abstract Present study develops short

More information

Real-Time Travel Time Prediction Using Multi-level k-nearest Neighbor Algorithm and Data Fusion Method

Real-Time Travel Time Prediction Using Multi-level k-nearest Neighbor Algorithm and Data Fusion Method 1861 Real-Time Travel Time Prediction Using Multi-level k-nearest Neighbor Algorithm and Data Fusion Method Sehyun Tak 1, Sunghoon Kim 2, Kiate Jang 3 and Hwasoo Yeo 4 1 Smart Transportation System Laboratory,

More information

Serious limitations of (single-layer) perceptrons: Cannot learn non-linearly separable tasks. Cannot approximate (learn) non-linear functions

Serious limitations of (single-layer) perceptrons: Cannot learn non-linearly separable tasks. Cannot approximate (learn) non-linear functions BACK-PROPAGATION NETWORKS Serious limitations of (single-layer) perceptrons: Cannot learn non-linearly separable tasks Cannot approximate (learn) non-linear functions Difficult (if not impossible) to design

More information

Unit III. A Survey of Neural Network Model

Unit III. A Survey of Neural Network Model Unit III A Survey of Neural Network Model 1 Single Layer Perceptron Perceptron the first adaptive network architecture was invented by Frank Rosenblatt in 1957. It can be used for the classification of

More information

Data and prognosis for renewable energy

Data and prognosis for renewable energy The Hong Kong Polytechnic University Department of Electrical Engineering Project code: FYP_27 Data and prognosis for renewable energy by Choi Man Hin 14072258D Final Report Bachelor of Engineering (Honours)

More information

Hurricane Harvey and What is to come. Eric Berger, Space City Weather

Hurricane Harvey and What is to come. Eric Berger, Space City Weather Hurricane Harvey and What is to come Eric Berger, Space City Weather Today s talk A word about Space City Weather Social media-rology A look back at Hurricane Harvey Why does Houston get these storms Rainfall

More information

Artificial Neural Network

Artificial Neural Network Artificial Neural Network Contents 2 What is ANN? Biological Neuron Structure of Neuron Types of Neuron Models of Neuron Analogy with human NN Perceptron OCR Multilayer Neural Network Back propagation

More information

The Research of Urban Rail Transit Sectional Passenger Flow Prediction Method

The Research of Urban Rail Transit Sectional Passenger Flow Prediction Method Journal of Intelligent Learning Systems and Applications, 2013, 5, 227-231 Published Online November 2013 (http://www.scirp.org/journal/jilsa) http://dx.doi.org/10.4236/jilsa.2013.54026 227 The Research

More information

Application of Artificial Neural Networks in Evaluation and Identification of Electrical Loss in Transformers According to the Energy Consumption

Application of Artificial Neural Networks in Evaluation and Identification of Electrical Loss in Transformers According to the Energy Consumption Application of Artificial Neural Networks in Evaluation and Identification of Electrical Loss in Transformers According to the Energy Consumption ANDRÉ NUNES DE SOUZA, JOSÉ ALFREDO C. ULSON, IVAN NUNES

More information

Transportation and Road Weather

Transportation and Road Weather Portland State University PDXScholar TREC Friday Seminar Series Transportation Research and Education Center (TREC) 4-18-2014 Transportation and Road Weather Rhonda Young University of Wyoming Let us know

More information

Neural Networks biological neuron artificial neuron 1

Neural Networks biological neuron artificial neuron 1 Neural Networks biological neuron artificial neuron 1 A two-layer neural network Output layer (activation represents classification) Weighted connections Hidden layer ( internal representation ) Input

More information

WHITE PAPER. Winter Tires

WHITE PAPER. Winter Tires WHITE PAPER Winter Tires WHITE PAPER Introduction According to the Federal Highway Administration accidents caused by winter weather result in 150,000 injuries and 2,000 deaths each year on average. In

More information

Appendix BAL Baltimore, Maryland 2003 Annual Report on Freeway Mobility and Reliability

Appendix BAL Baltimore, Maryland 2003 Annual Report on Freeway Mobility and Reliability (http://mobility.tamu.edu/mmp) Office of Operations, Federal Highway Administration Appendix BAL Baltimore, Maryland 2003 Annual Report on Freeway Mobility and Reliability This report is a supplement to:

More information

AN ARTIFICIAL NEURAL NETWORK MODEL FOR ROAD ACCIDENT PREDICTION: A CASE STUDY OF KHULNA METROPOLITAN CITY

AN ARTIFICIAL NEURAL NETWORK MODEL FOR ROAD ACCIDENT PREDICTION: A CASE STUDY OF KHULNA METROPOLITAN CITY Proceedings of the 4 th International Conference on Civil Engineering for Sustainable Development (ICCESD 2018), 9~11 February 2018, KUET, Khulna, Bangladesh (ISBN-978-984-34-3502-6) AN ARTIFICIAL NEURAL

More information

Neural Networks and the Back-propagation Algorithm

Neural Networks and the Back-propagation Algorithm Neural Networks and the Back-propagation Algorithm Francisco S. Melo In these notes, we provide a brief overview of the main concepts concerning neural networks and the back-propagation algorithm. We closely

More information

y(x n, w) t n 2. (1)

y(x n, w) t n 2. (1) Network training: Training a neural network involves determining the weight parameter vector w that minimizes a cost function. Given a training set comprising a set of input vector {x n }, n = 1,...N,

More information

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others)

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others) Machine Learning Neural Networks (slides from Domingos, Pardo, others) For this week, Reading Chapter 4: Neural Networks (Mitchell, 1997) See Canvas For subsequent weeks: Scaling Learning Algorithms toward

More information

Short-term traffic volume prediction using neural networks

Short-term traffic volume prediction using neural networks Short-term traffic volume prediction using neural networks Nassim Sohaee Hiranya Garbha Kumar Aswin Sreeram Abstract There are many modeling techniques that can predict the behavior of complex systems,

More information

Classification with Perceptrons. Reading:

Classification with Perceptrons. Reading: Classification with Perceptrons Reading: Chapters 1-3 of Michael Nielsen's online book on neural networks covers the basics of perceptrons and multilayer neural networks We will cover material in Chapters

More information

Data Mining Part 5. Prediction

Data Mining Part 5. Prediction Data Mining Part 5. Prediction 5.5. Spring 2010 Instructor: Dr. Masoud Yaghini Outline How the Brain Works Artificial Neural Networks Simple Computing Elements Feed-Forward Networks Perceptrons (Single-layer,

More information

Use of Weather Inputs in Traffic Volume Forecasting

Use of Weather Inputs in Traffic Volume Forecasting ISSC 7, Derry, Sept 13 14 Use of Weather Inputs in Traffic Volume Forecasting Shane Butler, John Ringwood and Damien Fay Department of Electronic Engineering NUI Maynooth, Maynooth Co. Kildare, Ireland

More information

Multilayer Perceptrons (MLPs)

Multilayer Perceptrons (MLPs) CSE 5526: Introduction to Neural Networks Multilayer Perceptrons (MLPs) 1 Motivation Multilayer networks are more powerful than singlelayer nets Example: XOR problem x 2 1 AND x o x 1 x 2 +1-1 o x x 1-1

More information

Mr. Harshit K. Dave 1, Dr. Keyur P. Desai 2, Dr. Harit K. Raval 3

Mr. Harshit K. Dave 1, Dr. Keyur P. Desai 2, Dr. Harit K. Raval 3 Investigations on Prediction of MRR and Surface Roughness on Electro Discharge Machine Using Regression Analysis and Artificial Neural Network Programming Mr. Harshit K. Dave 1, Dr. Keyur P. Desai 2, Dr.

More information

EPL442: Computational

EPL442: Computational EPL442: Computational Learning Systems Lab 2 Vassilis Vassiliades Department of Computer Science University of Cyprus Outline Artificial Neuron Feedforward Neural Network Back-propagation Algorithm Notes

More information

CONGESTION REPORT 1 st Quarter 2016

CONGESTION REPORT 1 st Quarter 2016 CONGESTION REPORT 1 st Quarter 2016 A quarterly update of the National Capital Region s traffic congestion, travel time reliability, top-10 bottlenecks and featured spotlight April 20, 2016 ABOUT TPB Transportation

More information

Introduction to Machine Learning

Introduction to Machine Learning Introduction to Machine Learning Neural Networks Varun Chandola x x 5 Input Outline Contents February 2, 207 Extending Perceptrons 2 Multi Layered Perceptrons 2 2. Generalizing to Multiple Labels.................

More information

ANN and Statistical Theory Based Forecasting and Analysis of Power System Variables

ANN and Statistical Theory Based Forecasting and Analysis of Power System Variables ANN and Statistical Theory Based Forecasting and Analysis of Power System Variables Sruthi V. Nair 1, Poonam Kothari 2, Kushal Lodha 3 1,2,3 Lecturer, G. H. Raisoni Institute of Engineering & Technology,

More information

Weather Information for Road Managers

Weather Information for Road Managers Weather Information for Road Managers by Paul A. Pisano, Lynette C. Goodwin and Andrew D. Stern How Does Weather Affect Roads? The complex interactions between weather and roads have major affects on traffic

More information

Multilayer Perceptrons and Backpropagation

Multilayer Perceptrons and Backpropagation Multilayer Perceptrons and Backpropagation Informatics 1 CG: Lecture 7 Chris Lucas School of Informatics University of Edinburgh January 31, 2017 (Slides adapted from Mirella Lapata s.) 1 / 33 Reading:

More information

Introduction to Natural Computation. Lecture 9. Multilayer Perceptrons and Backpropagation. Peter Lewis

Introduction to Natural Computation. Lecture 9. Multilayer Perceptrons and Backpropagation. Peter Lewis Introduction to Natural Computation Lecture 9 Multilayer Perceptrons and Backpropagation Peter Lewis 1 / 25 Overview of the Lecture Why multilayer perceptrons? Some applications of multilayer perceptrons.

More information

Deploying the Winter Maintenance Support System (MDSS) in Iowa

Deploying the Winter Maintenance Support System (MDSS) in Iowa Deploying the Winter Maintenance Support System (MDSS) in Iowa Dennis A. Kroeger Center for Transportation Research and Education Iowa State University Ames, IA 50010-8632 kroeger@iastate.edu Dennis Burkheimer

More information

Multilayer Neural Networks. (sometimes called Multilayer Perceptrons or MLPs)

Multilayer Neural Networks. (sometimes called Multilayer Perceptrons or MLPs) Multilayer Neural Networks (sometimes called Multilayer Perceptrons or MLPs) Linear separability Hyperplane In 2D: w x + w 2 x 2 + w 0 = 0 Feature x 2 = w w 2 x w 0 w 2 Feature 2 A perceptron can separate

More information

FHWA Road Weather Management Program Update

FHWA Road Weather Management Program Update FHWA Road Weather Management Program Update 2015 Winter Maintenance Peer Exchange Bloomington, MN September 21-25, 2015 Gabe Guevara FHWA Office of Operations Road Weather Management Team 2015 Winter Maintenance

More information

Multilayer Neural Networks. (sometimes called Multilayer Perceptrons or MLPs)

Multilayer Neural Networks. (sometimes called Multilayer Perceptrons or MLPs) Multilayer Neural Networks (sometimes called Multilayer Perceptrons or MLPs) Linear separability Hyperplane In 2D: w 1 x 1 + w 2 x 2 + w 0 = 0 Feature 1 x 2 = w 1 w 2 x 1 w 0 w 2 Feature 2 A perceptron

More information

Forecasting of Rain Fall in Mirzapur District, Uttar Pradesh, India Using Feed-Forward Artificial Neural Network

Forecasting of Rain Fall in Mirzapur District, Uttar Pradesh, India Using Feed-Forward Artificial Neural Network International Journal of Engineering Science Invention ISSN (Online): 2319 6734, ISSN (Print): 2319 6726 Volume 2 Issue 8ǁ August. 2013 ǁ PP.87-93 Forecasting of Rain Fall in Mirzapur District, Uttar Pradesh,

More information

Responsive Traffic Management Through Short-Term Weather and Collision Prediction

Responsive Traffic Management Through Short-Term Weather and Collision Prediction Responsive Traffic Management Through Short-Term Weather and Collision Prediction Presenter: Stevanus A. Tjandra, Ph.D. City of Edmonton Office of Traffic Safety (OTS) Co-authors: Yongsheng Chen, Ph.D.,

More information

Neural Networks. CSE 6363 Machine Learning Vassilis Athitsos Computer Science and Engineering Department University of Texas at Arlington

Neural Networks. CSE 6363 Machine Learning Vassilis Athitsos Computer Science and Engineering Department University of Texas at Arlington Neural Networks CSE 6363 Machine Learning Vassilis Athitsos Computer Science and Engineering Department University of Texas at Arlington 1 Perceptrons x 0 = 1 x 1 x 2 z = h w T x Output: z x D A perceptron

More information

Predicting flight on-time performance

Predicting flight on-time performance 1 Predicting flight on-time performance Arjun Mathur, Aaron Nagao, Kenny Ng I. INTRODUCTION Time is money, and delayed flights are a frequent cause of frustration for both travellers and airline companies.

More information

Last update: October 26, Neural networks. CMSC 421: Section Dana Nau

Last update: October 26, Neural networks. CMSC 421: Section Dana Nau Last update: October 26, 207 Neural networks CMSC 42: Section 8.7 Dana Nau Outline Applications of neural networks Brains Neural network units Perceptrons Multilayer perceptrons 2 Example Applications

More information

Artificial Neural Networks

Artificial Neural Networks Artificial Neural Networks Threshold units Gradient descent Multilayer networks Backpropagation Hidden layer representations Example: Face Recognition Advanced topics 1 Connectionist Models Consider humans:

More information

Revision: Neural Network

Revision: Neural Network Revision: Neural Network Exercise 1 Tell whether each of the following statements is true or false by checking the appropriate box. Statement True False a) A perceptron is guaranteed to perfectly learn

More information

Electric Load Forecasting Using Wavelet Transform and Extreme Learning Machine

Electric Load Forecasting Using Wavelet Transform and Extreme Learning Machine Electric Load Forecasting Using Wavelet Transform and Extreme Learning Machine Song Li 1, Peng Wang 1 and Lalit Goel 1 1 School of Electrical and Electronic Engineering Nanyang Technological University

More information

Module 12. Machine Learning. Version 2 CSE IIT, Kharagpur

Module 12. Machine Learning. Version 2 CSE IIT, Kharagpur Module 12 Machine Learning Lesson 39 Neural Networks - III 12.4.4 Multi-Layer Perceptrons In contrast to perceptrons, multilayer networks can learn not only multiple decision boundaries, but the boundaries

More information

Artificial Neural Networks Examination, June 2005

Artificial Neural Networks Examination, June 2005 Artificial Neural Networks Examination, June 2005 Instructions There are SIXTY questions. (The pass mark is 30 out of 60). For each question, please select a maximum of ONE of the given answers (either

More information

Classification goals: Make 1 guess about the label (Top-1 error) Make 5 guesses about the label (Top-5 error) No Bounding Box

Classification goals: Make 1 guess about the label (Top-1 error) Make 5 guesses about the label (Top-5 error) No Bounding Box ImageNet Classification with Deep Convolutional Neural Networks Alex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton Motivation Classification goals: Make 1 guess about the label (Top-1 error) Make 5 guesses

More information

An Interruption in the Highway: New Approach to Modeling Car Traffic

An Interruption in the Highway: New Approach to Modeling Car Traffic An Interruption in the Highway: New Approach to Modeling Car Traffic Amin Rezaeezadeh * Physics Department Sharif University of Technology Tehran, Iran Received: February 17, 2010 Accepted: February 9,

More information

Unit 8: Introduction to neural networks. Perceptrons

Unit 8: Introduction to neural networks. Perceptrons Unit 8: Introduction to neural networks. Perceptrons D. Balbontín Noval F. J. Martín Mateos J. L. Ruiz Reina A. Riscos Núñez Departamento de Ciencias de la Computación e Inteligencia Artificial Universidad

More information

Direct Method for Training Feed-forward Neural Networks using Batch Extended Kalman Filter for Multi- Step-Ahead Predictions

Direct Method for Training Feed-forward Neural Networks using Batch Extended Kalman Filter for Multi- Step-Ahead Predictions Direct Method for Training Feed-forward Neural Networks using Batch Extended Kalman Filter for Multi- Step-Ahead Predictions Artem Chernodub, Institute of Mathematical Machines and Systems NASU, Neurotechnologies

More information

Acquisition of Multi-Function Equipment at DIA: Conditions, Factors, Considerations & Integration. Presented by Mike Carlson September 20, 2012

Acquisition of Multi-Function Equipment at DIA: Conditions, Factors, Considerations & Integration. Presented by Mike Carlson September 20, 2012 Acquisition of Multi-Function Equipment at DIA: Conditions, Factors, Considerations & Integration Presented by Mike Carlson September 20, 2012 1 Denver International Airport 5 Runways 12,000 /. (3,658m)

More information

Administration. Registration Hw3 is out. Lecture Captioning (Extra-Credit) Scribing lectures. Questions. Due on Thursday 10/6

Administration. Registration Hw3 is out. Lecture Captioning (Extra-Credit) Scribing lectures. Questions. Due on Thursday 10/6 Administration Registration Hw3 is out Due on Thursday 10/6 Questions Lecture Captioning (Extra-Credit) Look at Piazza for details Scribing lectures With pay; come talk to me/send email. 1 Projects Projects

More information

Neural Networks Task Sheet 2. Due date: May

Neural Networks Task Sheet 2. Due date: May Neural Networks 2007 Task Sheet 2 1/6 University of Zurich Prof. Dr. Rolf Pfeifer, pfeifer@ifi.unizh.ch Department of Informatics, AI Lab Matej Hoffmann, hoffmann@ifi.unizh.ch Andreasstrasse 15 Marc Ziegler,

More information

Machine Learning. Neural Networks

Machine Learning. Neural Networks Machine Learning Neural Networks Bryan Pardo, Northwestern University, Machine Learning EECS 349 Fall 2007 Biological Analogy Bryan Pardo, Northwestern University, Machine Learning EECS 349 Fall 2007 THE

More information

Analysis of Fast Input Selection: Application in Time Series Prediction

Analysis of Fast Input Selection: Application in Time Series Prediction Analysis of Fast Input Selection: Application in Time Series Prediction Jarkko Tikka, Amaury Lendasse, and Jaakko Hollmén Helsinki University of Technology, Laboratory of Computer and Information Science,

More information

Predicting MTA Bus Arrival Times in New York City

Predicting MTA Bus Arrival Times in New York City Predicting MTA Bus Arrival Times in New York City Man Geen Harold Li, CS 229 Final Project ABSTRACT This final project sought to outperform the Metropolitan Transportation Authority s estimated time of

More information

AN INTRODUCTION TO NEURAL NETWORKS. Scott Kuindersma November 12, 2009

AN INTRODUCTION TO NEURAL NETWORKS. Scott Kuindersma November 12, 2009 AN INTRODUCTION TO NEURAL NETWORKS Scott Kuindersma November 12, 2009 SUPERVISED LEARNING We are given some training data: We must learn a function If y is discrete, we call it classification If it is

More information

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others)

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others) Machine Learning Neural Networks (slides from Domingos, Pardo, others) For this week, Reading Chapter 4: Neural Networks (Mitchell, 1997) See Canvas For subsequent weeks: Scaling Learning Algorithms toward

More information

Introduction To Artificial Neural Networks

Introduction To Artificial Neural Networks Introduction To Artificial Neural Networks Machine Learning Supervised circle square circle square Unsupervised group these into two categories Supervised Machine Learning Supervised Machine Learning Supervised

More information

ESTIMATING THE ACTIVATION FUNCTIONS OF AN MLP-NETWORK

ESTIMATING THE ACTIVATION FUNCTIONS OF AN MLP-NETWORK ESTIMATING THE ACTIVATION FUNCTIONS OF AN MLP-NETWORK P.V. Vehviläinen, H.A.T. Ihalainen Laboratory of Measurement and Information Technology Automation Department Tampere University of Technology, FIN-,

More information

1 What a Neural Network Computes

1 What a Neural Network Computes Neural Networks 1 What a Neural Network Computes To begin with, we will discuss fully connected feed-forward neural networks, also known as multilayer perceptrons. A feedforward neural network consists

More information

Computational statistics

Computational statistics Computational statistics Lecture 3: Neural networks Thierry Denœux 5 March, 2016 Neural networks A class of learning methods that was developed separately in different fields statistics and artificial

More information

Predicting freeway traffic in the Bay Area

Predicting freeway traffic in the Bay Area Predicting freeway traffic in the Bay Area Jacob Baldwin Email: jtb5np@stanford.edu Chen-Hsuan Sun Email: chsun@stanford.edu Ya-Ting Wang Email: yatingw@stanford.edu Abstract The hourly occupancy rate

More information

Internet Engineering Jacek Mazurkiewicz, PhD

Internet Engineering Jacek Mazurkiewicz, PhD Internet Engineering Jacek Mazurkiewicz, PhD Softcomputing Part 11: SoftComputing Used for Big Data Problems Agenda Climate Changes Prediction System Based on Weather Big Data Visualisation Natural Language

More information

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others)

Machine Learning. Neural Networks. (slides from Domingos, Pardo, others) Machine Learning Neural Networks (slides from Domingos, Pardo, others) Human Brain Neurons Input-Output Transformation Input Spikes Output Spike Spike (= a brief pulse) (Excitatory Post-Synaptic Potential)

More information

SOIL MOISTURE MODELING USING ARTIFICIAL NEURAL NETWORKS

SOIL MOISTURE MODELING USING ARTIFICIAL NEURAL NETWORKS Int'l Conf. Artificial Intelligence ICAI'17 241 SOIL MOISTURE MODELING USING ARTIFICIAL NEURAL NETWORKS Dr. Jayachander R. Gangasani Instructor, Department of Computer Science, jay.gangasani@aamu.edu Dr.

More information

U.S. - Canadian Border Traffic Prediction

U.S. - Canadian Border Traffic Prediction Western Washington University Western CEDAR WWU Honors Program Senior Projects WWU Graduate and Undergraduate Scholarship 12-14-2017 U.S. - Canadian Border Traffic Prediction Colin Middleton Western Washington

More information

Nonlinear System Identification Using MLP Dr.-Ing. Sudchai Boonto

Nonlinear System Identification Using MLP Dr.-Ing. Sudchai Boonto Dr-Ing Sudchai Boonto Department of Control System and Instrumentation Engineering King Mongkut s Unniversity of Technology Thonburi Thailand Nonlinear System Identification Given a data set Z N = {y(k),

More information

Organized Chain-Up and VSL

Organized Chain-Up and VSL Organized Chain-Up and VSL Jim Mahugh, PE WSDOT SC Region Traffic Engineer North/West Passage VSL Peer Exchange January 28, 2015 Snoqualmie Pass 2 Limits of VSL EB: MP 48.12 to 66.56 WB: MP 46.69 to 66.90

More information

CHAPTER 5 FORECASTING TRAVEL TIME WITH NEURAL NETWORKS

CHAPTER 5 FORECASTING TRAVEL TIME WITH NEURAL NETWORKS CHAPTER 5 FORECASTING TRAVEL TIME WITH NEURAL NETWORKS 5.1 - Introduction The estimation and predication of link travel times in a road traffic network are critical for many intelligent transportation

More information

SPSS, University of Texas at Arlington. Topics in Machine Learning-EE 5359 Neural Networks

SPSS, University of Texas at Arlington. Topics in Machine Learning-EE 5359 Neural Networks Topics in Machine Learning-EE 5359 Neural Networks 1 The Perceptron Output: A perceptron is a function that maps D-dimensional vectors to real numbers. For notational convenience, we add a zero-th dimension

More information

Pattern Recognition Prof. P. S. Sastry Department of Electronics and Communication Engineering Indian Institute of Science, Bangalore

Pattern Recognition Prof. P. S. Sastry Department of Electronics and Communication Engineering Indian Institute of Science, Bangalore Pattern Recognition Prof. P. S. Sastry Department of Electronics and Communication Engineering Indian Institute of Science, Bangalore Lecture - 27 Multilayer Feedforward Neural networks with Sigmoidal

More information

Accuracy improvement program for VMap1 to Multinational Geospatial Co-production Program (MGCP) using artificial neural networks

Accuracy improvement program for VMap1 to Multinational Geospatial Co-production Program (MGCP) using artificial neural networks 7th International Symposium on Spatial Accuracy Assessment in Natural Resources and Environmental Sciences. Edited by M. Caetano and M. Painho. Accuracy improvement program for VMap1 to Multinational Geospatial

More information

What Patterns Can Be Observed in a Year?

What Patterns Can Be Observed in a Year? LESSON 3 What Patterns Can Be Observed in a Year? From this vantage point, you can see the moon, sun, stars, and Earth. From Earth s surface, there are patterns to how the sun, moon, and stars appear in

More information

Introduction to Neural Networks

Introduction to Neural Networks Introduction to Neural Networks What are (Artificial) Neural Networks? Models of the brain and nervous system Highly parallel Process information much more like the brain than a serial computer Learning

More information

epochs epochs

epochs epochs Neural Network Experiments To illustrate practical techniques, I chose to use the glass dataset. This dataset has 214 examples and 6 classes. Here are 4 examples from the original dataset. The last values

More information

Complete Weather Intelligence for Public Safety from DTN

Complete Weather Intelligence for Public Safety from DTN Complete Weather Intelligence for Public Safety from DTN September 2017 White Paper www.dtn.com / 1.800.610.0777 From flooding to tornados to severe winter storms, the threats to public safety from weather-related

More information

DEVELOPMENT OF ROAD SURFACE TEMPERATURE PREDICTION MODEL

DEVELOPMENT OF ROAD SURFACE TEMPERATURE PREDICTION MODEL International Journal of Civil, Structural, Environmental and Infrastructure Engineering Research and Development (IJCSEIERD) ISSN(P): 2249-6866; ISSN(E): 2249-7978 Vol. 6, Issue 6, Dec 2016, 27-34 TJPRC

More information

Analyzing the effect of Weather on Uber Ridership

Analyzing the effect of Weather on Uber Ridership ABSTRACT MWSUG 2016 Paper AA22 Analyzing the effect of Weather on Uber Ridership Snigdha Gutha, Oklahoma State University Anusha Mamillapalli, Oklahoma State University Uber has changed the face of taxi

More information

ARTIFICIAL NEURAL NETWORK PART I HANIEH BORHANAZAD

ARTIFICIAL NEURAL NETWORK PART I HANIEH BORHANAZAD ARTIFICIAL NEURAL NETWORK PART I HANIEH BORHANAZAD WHAT IS A NEURAL NETWORK? The simplest definition of a neural network, more properly referred to as an 'artificial' neural network (ANN), is provided

More information

A Logarithmic Neural Network Architecture for Unbounded Non-Linear Function Approximation

A Logarithmic Neural Network Architecture for Unbounded Non-Linear Function Approximation 1 Introduction A Logarithmic Neural Network Architecture for Unbounded Non-Linear Function Approximation J Wesley Hines Nuclear Engineering Department The University of Tennessee Knoxville, Tennessee,

More information

MDSS and Advanced Forecasting STSMO Weather Workshop Dave Huft, SDDOT Research Program Manager September 12, 2017

MDSS and Advanced Forecasting STSMO Weather Workshop Dave Huft, SDDOT Research Program Manager September 12, 2017 Connecting South Dakota and the Nation MDSS and Advanced Forecasting STSMO Weather Workshop Dave Huft, SDDOT Research Program Manager September 12, 2017 1 Weather Impacts & Strategies Winter perceived

More information

Annual Collision Report

Annual Collision Report 2016 Annual Collision Report Contents The Annual Collision Report is a summary of statistics associated with traffic collisions that occurred in the City of Winnipeg. This information is provided by Manitoba

More information

Neural networks. Chapter 19, Sections 1 5 1

Neural networks. Chapter 19, Sections 1 5 1 Neural networks Chapter 19, Sections 1 5 Chapter 19, Sections 1 5 1 Outline Brains Neural networks Perceptrons Multilayer perceptrons Applications of neural networks Chapter 19, Sections 1 5 2 Brains 10

More information

Winter Weather Get It Together. Infographic Social Media Campaign Transcomm 2018 Awards - 8a: Logo, Illustration and Infographic

Winter Weather Get It Together. Infographic Social Media Campaign Transcomm 2018 Awards - 8a: Logo, Illustration and Infographic Winter Weather Get It Together Infographic Social Media Campaign Transcomm 2018 Awards - 8a: Logo, Illustration and Infographic Background: Every winter, snow, ice and cold present a challenge for motorists

More information

USDOT Applications Utilizing RWIS

USDOT Applications Utilizing RWIS USDOT Applications Utilizing RWIS Benjamin McKeever, P.E. Program Manager, Traveler Info & Road Weather Research & Innovative Technology Administration U.S. Department of Transportation September 4, 2008

More information

Research Note INTELLIGENT FORECASTING OF RAINFALL AND TEMPERATURE OF SHIRAZ CITY USING NEURAL NETWORKS *

Research Note INTELLIGENT FORECASTING OF RAINFALL AND TEMPERATURE OF SHIRAZ CITY USING NEURAL NETWORKS * Iranian Journal of Science & Technology, Transaction B, Vol. 28, No. B1 Printed in Islamic Republic of Iran, 24 Shiraz University Research Note INTELLIGENT FORECASTING OF RAINFALL AND TEMPERATURE OF SHIRAZ

More information

Portugaliae Electrochimica Acta 26/4 (2008)

Portugaliae Electrochimica Acta 26/4 (2008) Portugaliae Electrochimica Acta 6/4 (008) 6-68 PORTUGALIAE ELECTROCHIMICA ACTA Comparison of Regression Model and Artificial Neural Network Model for the Prediction of Volume Percent of Diamond Deposition

More information

CSE446: Neural Networks Spring Many slides are adapted from Carlos Guestrin and Luke Zettlemoyer

CSE446: Neural Networks Spring Many slides are adapted from Carlos Guestrin and Luke Zettlemoyer CSE446: Neural Networks Spring 2017 Many slides are adapted from Carlos Guestrin and Luke Zettlemoyer Human Neurons Switching time ~ 0.001 second Number of neurons 10 10 Connections per neuron 10 4-5 Scene

More information

Artificial Intelligence

Artificial Intelligence Artificial Intelligence Jeff Clune Assistant Professor Evolving Artificial Intelligence Laboratory Announcements Be making progress on your projects! Three Types of Learning Unsupervised Supervised Reinforcement

More information

Prediction for night-time ventilation in Stanford s Y2E2 building

Prediction for night-time ventilation in Stanford s Y2E2 building Prediction for night-time ventilation in Stanford s Y2E2 building Balthazar Donon Stanford University December 16, 2016 Indoor ventilation I. Introduction In the United States, around 40% of the energy

More information

Comparison of the Complex Valued and Real Valued Neural Networks Trained with Gradient Descent and Random Search Algorithms

Comparison of the Complex Valued and Real Valued Neural Networks Trained with Gradient Descent and Random Search Algorithms Comparison of the Complex Valued and Real Valued Neural Networks rained with Gradient Descent and Random Search Algorithms Hans Georg Zimmermann, Alexey Minin 2,3 and Victoria Kusherbaeva 3 - Siemens AG

More information

Road Weather Management Program

Road Weather Management Program Road Weather Management Program AASHTO/TRB Joint Maintenance Conference: Highway Safety & Reliability (Winter Maintenance and Highway Safety and Reliability) 7/22/2015 Gabe Guevara. P.E. FHWA Office of

More information

Comprehensive Winter Maintenance Management System BORRMA-web MDSS inside to increase Road Safety and Traffic Flow

Comprehensive Winter Maintenance Management System BORRMA-web MDSS inside to increase Road Safety and Traffic Flow Thorsten Cypra 1 Comprehensive Winter Maintenance Management System BORRMA-web MDSS inside to increase Road Safety and Traffic Flow American Public Works Association (APWA) Monday, 04/14/2008 3:30 4:30

More information

Analysis of Multilayer Neural Network Modeling and Long Short-Term Memory

Analysis of Multilayer Neural Network Modeling and Long Short-Term Memory Analysis of Multilayer Neural Network Modeling and Long Short-Term Memory Danilo López, Nelson Vera, Luis Pedraza International Science Index, Mathematical and Computational Sciences waset.org/publication/10006216

More information

Weather and Travel Time Decision Support

Weather and Travel Time Decision Support Weather and Travel Time Decision Support Gerry Wiener, Amanda Anderson, Seth Linden, Bill Petzke, Padhrig McCarthy, James Cowie, Thomas Brummet, Gabriel Guevara, Brenda Boyce, John Williams, Weiyan Chen

More information

DEEP LEARNING AND NEURAL NETWORKS: BACKGROUND AND HISTORY

DEEP LEARNING AND NEURAL NETWORKS: BACKGROUND AND HISTORY DEEP LEARNING AND NEURAL NETWORKS: BACKGROUND AND HISTORY 1 On-line Resources http://neuralnetworksanddeeplearning.com/index.html Online book by Michael Nielsen http://matlabtricks.com/post-5/3x3-convolution-kernelswith-online-demo

More information

CHALLENGE #1: ROAD CONDITIONS

CHALLENGE #1: ROAD CONDITIONS CHALLENGE #1: ROAD CONDITIONS Your forward collision warning system may struggle on wet or icy roads because it is not able to adjust for road conditions. Wet or slick roads may increase your stopping

More information

Understand FORWARD COLLISION WARNING WHAT IS IT? HOW DOES IT WORK? HOW TO USE IT?

Understand FORWARD COLLISION WARNING WHAT IS IT? HOW DOES IT WORK? HOW TO USE IT? Understand WHAT IS IT? Forward collision warning systems warn you of an impending collision by detecting stopped or slowly moved vehicles ahead of your vehicle. Forward collision warning use radar, lasers,

More information

WRTM and Active Transportation and Demand Management

WRTM and Active Transportation and Demand Management The Future of Transportation - 2010 APWA Annual Congress and Exposition WRTM and Active Transportation and Demand Management Weather Responsive Traffic Management Workshop October 7, 2011 Jim Hunt F d

More information

Lecture 5: Logistic Regression. Neural Networks

Lecture 5: Logistic Regression. Neural Networks Lecture 5: Logistic Regression. Neural Networks Logistic regression Comparison with generative models Feed-forward neural networks Backpropagation Tricks for training neural networks COMP-652, Lecture

More information