Analysis of Group Coding of Multiple Amino Acids in Artificial Neural Network Applied to the Prediction of Protein Secondary Structure

Size: px
Start display at page:

Download "Analysis of Group Coding of Multiple Amino Acids in Artificial Neural Network Applied to the Prediction of Protein Secondary Structure"

Transcription

1 Analysis of Grou Coding of Multile Amino Acids in Artificial Neural Networ Alied to the Prediction of Protein Secondary Structure Zhu Hong-ie 1, Dai Bin 2, Zhang Ya-feng 1, Bao Jia-li 3,* 1 College of Life Science, 2 Deartment of Mathematics, 3 College of Medicine, Zheiang University, Hangzhou, , China, * To whom corresondence should be addressed. Tel: ; baol@zu.edu.cn ABSTRACT In this aer, a new method involving grou coding is introduced into Artificial Neural Networ in the rediction of the secondary structure of the roteins. This method, to a larger degree, tae advantage of information of amino acids grous which ossibly lays a significant role in determining the secondary structure of the articular osition. Exeriments are conducted to test the efficiency of this method. The result shows that the rediction accuracy is significantly imroved, comared to the former method using single residue coding. We further discuss the mechanism of grou coding in the model and discuss its further imrovement in secondary rediction and other significant fields on structure and function rediction of bio-macromolecule based on sequence analysis. KEY WORDS: grou coding; Artificial Neural Networ (ANN); secondary structure rediction INTRODUCTION Since Qian and Senowshi [1] successfully alied this imlement to the field of secondary structure rediction, imrovement of the basic method mushrooms, including four resects: (1) Coding mode, rofile coding mode can incororate evolution information into the model, which imroves the rediction accuracy. [2] (2) Structure of the model, several identical ANN are trained on the same data set, then the rediction results are combined, which can bring about higher accuracy [3]. Additionally, it structurally imroves the connecting weights-sharing of model erformance, so revents the over-fitting roblem effectively [5]. (3) Otimization of the training rocess, dynamic factors, dynamic adustment of train rate is used to accelerate the seed of convergence of the training rocess. (4) Integration of other algorithms to Neural Networ. Genetic Algorithm [4] and Bayesian Model have been integrated to ANN to imrove the rediction accuracy. Most of these methods merely consider the rotein molecule as discrete symbol sequence. Contiguous residues in sequences are studied as a whole unit, which taes fully advantage of information hiding in amino acids grou. This method is called grou coding. Grou coding exresses couling effect of olyetide, which lays a significant role in determining the secondary structure of the articular osition. In this aer, couling effect of amino acids grous are taen into analysis in the ANN model; different inds of grou coding is comared, and efficiency of this method is tested. The result shows that the rediction accuracy is significantly imroved. In addition, the mechanism of the grou coding in the model is exlained; its further imrovement as well as other significant fields on structure and function rediction of biomacromolecule based on sequence analysis is discussed. MATERIAL AND METHODS Data The rotein data used in the study were obtained from Broohaven Protein Databan (PDB), and the data sets, within which the redundant sequences were excluded, included 72 roteins which belong to the family of rotein inase-lie (PK-lie) according to the classification of the Structural Classification of Proteins (SCOP) database. Total amount of residuals is while the number of

2 Helix, Sheet and Coil was resectively 10311(41.5%), 3638(14.7%) and 10876(43.8%). Artificial Neural Networ Model Predicting secondary structure of rotein by artificial neural networ contains two rocesses training and redicting. In the training rocess the structure of ANN model is firstly determined, and the arameters are then modified using training set, finally the ANN model constructed. The structure of the model is shown in fig.1, which is comosed by inut layer I, hidden layer J and outut layer K. There are 17 units in the inut layer, 40 in hidden layer whose inut value is denoted as x i and outut value as x, and 3 in outut layer whose outut value is x. The connecting weight from inut unit i to hidden unit is denoted as ω i, and that from hidden unit and outut unit is ω. Helix Sheet Coil Outut Vector K Outut Layer J Hidden Layer I Inut Layer Fig.1 toologic architecture of neural networ Inut window The inut layer of the neural networ is a window consisted of 17 units each of which is called a osition, containing one of the 20 amino acids or sace. A osition is actually an array comosed by 21 units, each secially denoting one ind of amino acids or sace. Therefore, the unit in a osition an array is given 1 if and only if it is consist with the containment of the osition or is given 0. This coding strategy is called single residue coding as is shown in figure 2. Fig.2 The structure of inut window and osition array The ninth osition of the inut window is the center one. An inut window with one residue as its center osition is inut in the ANN shown in fig.1. The three units of the outut layer stands for α Helix (H), β Sheet (E) and Coil resectively. The secondary structure whose corresonding outut unit has the largest outut value is considered as the one of the center osition, which is also called winner-tae-all. Inut coding of the osition array Grou coding codes two or more contiguous ositions in an inut window. The dietide code

3 consists of two ositions, each of which contains 21 osition units; therefore there are 21 2 coding tyes. For each tye, only one unit is 1, and the rest are 0. The trietide code consists of three ositions, each of which contains 21 osition units, therefore there are 21 3 coding tyes. For each tye, only one unit is 1, and the rest are 0. In a window with 17 ositions, there are 16 dietide codes and 15 trietide ones. Combination of the osition array To imrove the rediction accuracy, the toology structure of the ANN could be modified, which is achieved by combination of different inds of grou coding. As is shown in fig.1, within an inut window, there are 17 ositions for single residue coding, 16 for dietide coding and 15 for trietide coding. When single residue and dietide coding combine, for a single inut window, all units of the 17 ositions of single residue coding and 16 ositions of dietide coding construct the inut layer together. Similarly, the combination of dietide and trietide coding, single residue and trietide coding, furthermore all the three coding method could be taen into analysis in the same way, as is shown in tab.1. Table 1 Seven different combination methods of the osition array Denotation M D T MD DT MT MDT Single Bieide Trieide Cross-validation Due to the fact that results varied with the roteins selected for the testing and training sets, the cross-validation method is adoted to overcome the roblem in which the 72 rotein data were divided into 7 grous, two with 11 roteins and five with 10 ones. In the oulation of the 7 grous, each of them served as the testing set in turn, while the rest of the subgrous served as training sets. Therefore, 7 rounds of test roceeded. Afterwards the average of the 7 results was calculated to estimate the overall rediction accuracy for each exeriment [5]. Measurements of accuracy In this study, the rediction accuracy of rotein secondary structure is defined as follow: Rhelix + RSheet + RCoil Q3 = (1) N Here, R helix is the number of correctly redicted residues of α Helix, R sheet is the number of correctly redicted residues of β Sheet and R coil is the number of correctly redicted residues of Coil. N is total number of residues. In addition, the correlation coefficients of α Helix could also be calculated as: C H = ( n ) ( u o ) H H H H ( n + u )( n + o )( + u )( + o ) H H H H H H H H In the equation, H is the number of correctly redicted α Helix, n H is the number of residues that are correctly identified as something other than helix, o H is the number of nonhelical residues that are redicted as helix, and u H is the number of helical residues that are missed by the algorithm. A similar calculation could be easily alied to the correlation coefficients of β Sheet C E and Coil C C. BP Artificial Neural Networ Model Neural units Neural units are the fundamental comonents constructing the neural networ. The total inut of the unit is the sum of all the roduct of the inut x i and the connecting weightsω i : (2)

4 net = ω ixi i The outut x of the unit is determined by stimulating function: x = f (net ) (4) In this aer Sigmoid function is adoted as the stimulating function: 1 f (net ) = ( ) (5) net + θ / θ0 1 + e The arameter θ 0 mainly determines the shae of Sigmoid function withθ being the threshold. The outut of this unit x is a real number between 0 and 1. Exected function In the training rocess of the neural networ, a set of samles are rovided to be trained, in which the exected outut of the th outut unit for every samle is given as t. When a grou of connecting weights are given denoted as ω i and ω,the actual outut x could be calculated. Usually, exected outut t is not equal to real outut y, with error: E = 1 2 ( t x ) (6) 2 (3) For the whole set of the samles, its error is: 1 E = ( t x ) 2 2 (7) Increase of the connecting weights The increase of the connecting weights is calculated by bac-forward roagation algorithm (BP algorithm). When a grou of connecting weights ω i and ω are initially given, the networ estimates the difference between the actual outut y and the exected outut t based on the exected function. If the difference does not reach the exected function E 0, there would be the changes Δω i and Δω to all the connecting weights ω i and ω. The rocess will be reeated until E reaches the exected E 0. The training rocess coordinates the changes of connecting weights Δω i 和 Δω according to the rule that E is decreasing fastest. Moreover BP algorithm firstly calculates the changes of the inut layer Δω i, and then the changes of the outut layer Δω i according to gradient decent method in order to minimize the variation of the whole training set. Changes of the connecting weights in outut layer According to gradient decent training method, the changes of connecting weights in outut layer K for every samle are: Δω = η E = η in which η is training efficiency, from (3): = From (6) and (4): ω x = x (9) = δ = ( t x ) = (t x ) f (net ) (10) (8)

5 Consequently, tae (9) and (10) to (8),changes of the connecting weights of the unit in outut layer K are: Δω = η (t x ) f (net )x (11) Changes of connecting weights in hidden layer Changes of connecting weights in hidden layer J are: Δω i = η E i = η in which η is training efficiency, from (3): i i (12) = ω i xi = x i (13) i i From (6) and (4): = = f (net ) = f (net ) (14) According to (14): = ω x = ω (15) Tae (15) and (10) to (14), then (14) and (13) to (12), the changes of the connecting weights for a unit in hidden J are: Δω i = ηδ ω f (net ) x i (16) As is shown by (16), BP algorithm first calculates Δω in the outut layer, obtainingδ, and then Δω i. This algorithm achieves the minimization of the networ s error through gradient decent, imlementing training rocess according to the nown outut results of the networ. By adusting free arameters such as the number of hidden units, the number of training samles, the degree of homology between the training and testing sets, the length of inut window of residues, the algorithm searches the error sace and tries to find the local minimum, which yields the best classification for the training samles. RESULTS From the outcome shown in Tab.2, it is reasonable to draw the following conclusions: (1) When dietide (D) or trietide (T) serve as the inut unit for the model, the rediction accuracy is 9% and 11.1% higher than that of the model with single residue (M) as inut unit resectively, the coefficients of the three secondary structures are imroved too, which shows that grou coding is effective for imroving rediction accuracy and coefficient, and the trietide model is better then the dietide model. (2) The model with single residue and dietide (MD) as inut unit gives rediction accuracy 8.7% lower than that with dietide (D) as inut unit. As for the model with trietide (T) as inut unit, when single residue is incororated (MT) as inut unit, the corresonding rediction accuracy declines 3.9%. Similarly, coefficients also decline. Considered from model structure, when single residue is incororated as inut unit, it functions comaratively with dietide and trietide, therefore, affects the outut and rediction results. What deserves mentioning is that the decline of accuracy and coefficients is most severe for the model with dietide as inut unit.

6 (3) When dietide and trietide (DT) serve as inut unit simultaneously, the corresonding rediction accuracy culminates as high as 64.3%, meanwhile, coefficients also have maximums. However, after single residue unit is incororated, rediction accuracy declines 6.2%, and the corresonding coefficients decline corresondently, which shows that information rovided by dietide unit and trietide unit is comlementary to some extent, while information rovided by single residue contradicts with that of the former two sometimes. Table 2 The Prediction Accuracy of Seven Models M D T MD MT DT MDT 52.2% 61.2% 63.3% 52.5% 59.4% 64.3% 58.1% C H C S C C *Note: The numbers in the rows of Helix, Sheet and coil reresent corresonding correlation coefficients. DISCUSSION change of connecting weight 20 residues of are selected from the amino acid residues of the 72 roteins randomly. The secondary structure corresonding to the 20 residues is: KHHKKKHKHHKKHHKEHHKK (The symbol H stands for αhelix; K stands for all the other inds of secondary structure excet for helix). Helix and other inds of structure aear alternately for ten times. The change of the 40 weights belonging to the unit corresonding to helix in the outut layer between the situation after the first training rocess and that after the last training rocess. The figure informs that excet for the 35 th weights which exhibits oosite feature, the others commonly tends to ee the same feature as that formed in the first few iterations, In other words, the initial larger weights always remain larger than initial smaller ones, what is more, the difference between weights are magnified. It could therefore be concluded that the attern of weights in outut layer forms in the initial stages of training rocess and in the later course; the attern would not change dramatically. This situation would result in that weight attern of hidden layer exerts maor function in the erformance of the whole model. Because the introduction of grou coding mainly changes the toology structure of inut and hidden layers, the outut layer would not diminish its effect. The comare aforementioned after training from the first osition. It is evident that the oosition of atterns of the two units has aeared at this time which means that if a weight in one unit is resectively larger, its counterart in the other unit tends to be smaller and vice versa. This trend is more significant after the training from the 20 ositions. The secondary structure corresonding to the first osition is coil. After the first training from the first osition, the weight attern of outut layer corresonding to helix is basically reversed to the outut attern of hidden layer. By contrast, the situation changed dramatically when the target of training is changed to Helix in the training from the second osition. In other words, the two aforementioned atterns are basically consistent with each other. The same analysis was conducted to the following 18 training rocesses, and the result is similar with aforementioned ones. The training rocess thus could be inferred from the above analyze. For every single unit in outut layer, when the ind of structure it stands for is consistent with training target (this ind of unit is termed as this first ind of units in outut layer while the other two are termed as the second ind of units.), its connecting weight attern would be consistent with outut attern of hidden layer which means larger outut would be multilied by larger connecting weight. This attern of weighted sum

7 would contribute to larger outut of this articular unit in outut layer. On the contrary, for the other two units, the situation is reverse. According to the winner-tae-all rincile, the articular ind of secondary structure that the first ind of unit in outut layer stands for would be the result of final rediction. Once the connecting weight attern of outut layer is formed (the attern is originally formed according to the randomization of connecting weight in hidden layer), this attern will remain unchanged basically. Furthermore,it will shae the connecting weight attern in hidden layer in the following training rocess and further shae its outut attern. Hidden units are the basic unit of this rocess, which means all the connecting weights in one unit will increase or decrease in the same manner. The final result is reresented by the significant value difference between the oututs of hidden units. This is similar to a ind of encoding strategy. The algorithm of training rocess could exlain the aforementioned extraolation. Firstly, based on eq.15 and eq.16, for connecting weights belonging to the same hidden unit, the differences among their change Δω i only exist in the inut x i. However, the inut to hidden layer in this model is only 0 or 1. The inut unit whose value is 0 and their corresonding connecting weight in hidden layer is not taen into account. Therefore, Δω i for other connecting weight belonging to the same hidden unit would be identical which demonstrates that all the connecting weight in one unit will change in the same manner. Secondly, for a hidden unit, besides x i mentioned above, the change Δω i is determined by two searated arts f (net ) and δ ω. Since connecting weights are initialized randomly, the difference of t f (net ) for all the units could be considered as no significant. The second art is δ ( ω). When being initialized, it is determined by connecting weight and other values of outut layer. In this articular model, this art could be further divided into three subarts, corresonding to the three units in outut layer. As to the first ind of unit in outut layer, δ is ositive. For larger connecting weights, the δ ω they contribute would be larger. On the other hand, their counterarts in the other two units are resectively smaller. Nevertheless, since the δ of these two units is negative, the δ ω they contribute would be still resectively larger than they do to other hidden units. This interesting strategy would finally lead to the common increase of the connecting weights belonging to this articular hidden unit, which further contribute to its larger outut. As to the former art of f (net ) in the same training rocess, since net continually increases, it would gradually decrease in the later iterations and finally converges. To the oosite, if the corresonding connecting weight of the hidden unit U in the first ind of unit in outut layer is smaller, its connecting weights would be continually decreased which leads to its final smaller outut. With this strategy, the connecting weight attern of outut layer shaes the connecting weight attern in hidden layer and further shae its outut attern. In the other resect, as is shown in eq.11, for a unit in outut layer, the differences of the change Δω among its connecting weights only exist in x which is the outut of hidden layer. According to the connecting weight attern of outut layer, the connecting weights of the first ind of units could be categorized into two sets: the set of larger ones and the set of smaller ones. The outut of units in hidden layer corresonding to the first set of connecting weights would be significantly increased in the first modification rocess, and as a matter of which, they would also significantly increase. On the other hand, since the outut of units in hidden layer corresonding to the second set of connecting

8 weights would be decreased in the first modification rocess, though they would also be increased, yet the amlitude would be much smaller than the first set. Similarly, the connecting weights of the second ind of units could be categorized into two sets: the set of larger ones and the set of smaller ones. The outut of units in hidden layer corresonding to the first set of connecting weights would be decreased in the first modification rocess, though they would be decreased, yet the amlitude would not be exected large. The outut of units in hidden layer corresonding to the second set of connecting weights would be significantly increased, and as a matter of which, they would be significantly decreased. With this strategy, outut layer maintains and further enhance its connecting weight attern. The roerty of indication and secification of amino acid grou In the training set, if some amino acids grou is corresondent to a articular secondary structure with relatively higher robability, the air of the amino acids grou and the secondary structure, is called a high frequency. For examle, if an amino acid grou AAA always corresonds to α Helix, the air of AAA and H ( AAA, H) is called a high frequency and the airs ( AAA, C) and ( AAA, S) is called low frequency. If some amino acids grou in training set is never corresonded to a secondary structure, the air of them is called a null frequency. The ercentage of high frequency in the roteins being redicted is highly corresondent to the level of the rediction accuracy which reflecting the close relationshi between secific amino acid grou and secondary structure. This articular roerty of amino acid grou is called the roerty of indication. Furthermore, null frequencies hardly aear in the roteins being redicted, which means most of the airs of amino acids grou and secondary structure being redicted have aeared in the training rocess. This condition as well as the roerty of indication is combined and commonly called the roerty of secification of amino acid grou. The roerty of secification of amino acids grou can be exlained by the function of every amino acids hysical and chemical roerties to determine the secondary structure, including the size and olarity of R grous, hydrohobic and hydrohilic ability. Different residues have different tendencies to form αhelix and βsheet, therefore, a articular etide is rone to form a certain secondary structure while others are liely to form a different one [6]. For illustrations, olyisoleucine whose R grous are large enough to form satial barrier can not formαhelix. Similarly, Proline has a distinctive cyclic structure. As a result, when roline aears in olyetide, αhelix will be broe [7]. In addition, inβturn which is classified as coil here often aear olar residues such as asaragines, asartate, glutamine, serine and residues that affect the formation of αhelix andβsheet such as roline and glycine [6]. Integrating the above three arts, grou coding has certain function in imroving the rediction accuracy. In the rocess of training or rediction, because of the roerty of secification of amino acid grous, contiguous amino acids are more liely to aear in secific combinations and arrangements rather random manner. In terms of ANN model structure, it means that merely a small ortion of connecting weights are used and modified. The combinations of these connecting weights are more liely to reflect the atterns of the connecting weights of hidden layer, which can imrove the rediction accuracy remarably when combined with the roerty of indication of amino acid grous. If the connecting weight attern memorized in the training rocess can not reflect the actual structure of the rotein being redicted, the rediction ability of grou coding will be undermined, which means that the function of grou coding deends on homology of data, for examle the rediction of structure of function site. Because these ositions exerience more evolution ressure,

9 their sequence and their structure tend to be conserved. As for rediction of common structure, it is helful to do some homology analysis before redict secondary structure using grou coding. Besides, when several identical networs redict a single sequence, the rediction accuracy can be imroved effectively. Due to the fact that homology and training order of the samles exert rofound influence on the final rediction accuracy,a different multi-model networs method is used, in which the models are trained on different data set, or they are trained in different orders of samles within the same data set. The rediction results are accomlished resectively by all the models, and are combined to yield the final result. ACKNOWLEDGEMENTS The authors gratefully acnowledge the financial suort of Student Research Training Program of Zheiang University of China. REFERENCE [1] Ning Qian, Terrence J. Senowsi, Predicting the Secondary Structure of Globular Proteins Using Neural Networ Models. J. Mol. Biol.heiang, 202, [2] Ruan Xiao-gang, Sun Hai-un. Research on Encode Influencing Protein Secondary Structure Prediction. Journal of Beiing University of Technology. 31(3), , 2005 [3] Hanxi ZHU, Iuo YOSHIHARA, Kunihito YAMAMORI. Prediction of Protein Secondary Structure by Multi Model Neural Networs, Proc International Joint Conference on Neural Networs (IJCNN 02), , 2002 [4] A J F Van Rooi, RP Johnson, LC Jain, Neural Networ Training Using Genetic Algorithm, World Scientific, 1996 [5] Yan Hua-Jun, Fu Yan, Zhang Yi, Li Yi-Chao. Neural Networ Method for Protein Secondary Structure Prediction, Comuter science, 30(2), 48-52, 2003 [6] Yan Long-fei, Sun Zhi-rong. Moleculer Structure of Protein. Tsinghua University Press [7] Wang Jing-yan, Zhu Sheng-geng, Xu Chang-fa. Biochemistry. Higher Education Press. 2002

Feedback-error control

Feedback-error control Chater 4 Feedback-error control 4.1 Introduction This chater exlains the feedback-error (FBE) control scheme originally described by Kawato [, 87, 8]. FBE is a widely used neural network based controller

More information

Physicochemical properties of GPCR amino acid sequences for understanding GPCR-G-protein coupling

Physicochemical properties of GPCR amino acid sequences for understanding GPCR-G-protein coupling Chem-Bio Informatics Journal, Vol. 8, No. 2,.49-57 (2008) Physicochemical roerties of GPCR amino acid sequences for understanding GPCR-G-rotein couling Ganga D. Ghimire 1, 2*, Hideki Tanizawa 2, Masashi

More information

Multilayer Perceptron Neural Network (MLPs) For Analyzing the Properties of Jordan Oil Shale

Multilayer Perceptron Neural Network (MLPs) For Analyzing the Properties of Jordan Oil Shale World Alied Sciences Journal 5 (5): 546-552, 2008 ISSN 1818-4952 IDOSI Publications, 2008 Multilayer Percetron Neural Network (MLPs) For Analyzing the Proerties of Jordan Oil Shale 1 Jamal M. Nazzal, 2

More information

Research of power plant parameter based on the Principal Component Analysis method

Research of power plant parameter based on the Principal Component Analysis method Research of ower lant arameter based on the Princial Comonent Analysis method Yang Yang *a, Di Zhang b a b School of Engineering, Bohai University, Liaoning Jinzhou, 3; Liaoning Datang international Jinzhou

More information

Research Article Research on Evaluation Indicator System and Methods of Food Network Marketing Performance

Research Article Research on Evaluation Indicator System and Methods of Food Network Marketing Performance Advance Journal of Food Science and Technology 7(0: 80-84 205 DOI:0.9026/afst.7.988 ISSN: 2042-4868; e-issn: 2042-4876 205 Maxwell Scientific Publication Cor. Submitted: October 7 204 Acceted: December

More information

Metrics Performance Evaluation: Application to Face Recognition

Metrics Performance Evaluation: Application to Face Recognition Metrics Performance Evaluation: Alication to Face Recognition Naser Zaeri, Abeer AlSadeq, and Abdallah Cherri Electrical Engineering Det., Kuwait University, P.O. Box 5969, Safat 6, Kuwait {zaery, abeer,

More information

arxiv: v1 [physics.data-an] 26 Oct 2012

arxiv: v1 [physics.data-an] 26 Oct 2012 Constraints on Yield Parameters in Extended Maximum Likelihood Fits Till Moritz Karbach a, Maximilian Schlu b a TU Dortmund, Germany, moritz.karbach@cern.ch b TU Dortmund, Germany, maximilian.schlu@cern.ch

More information

Radial Basis Function Networks: Algorithms

Radial Basis Function Networks: Algorithms Radial Basis Function Networks: Algorithms Introduction to Neural Networks : Lecture 13 John A. Bullinaria, 2004 1. The RBF Maing 2. The RBF Network Architecture 3. Comutational Power of RBF Networks 4.

More information

Fuzzy Automata Induction using Construction Method

Fuzzy Automata Induction using Construction Method Journal of Mathematics and Statistics 2 (2): 395-4, 26 ISSN 1549-3644 26 Science Publications Fuzzy Automata Induction using Construction Method 1,2 Mo Zhi Wen and 2 Wan Min 1 Center of Intelligent Control

More information

Hotelling s Two- Sample T 2

Hotelling s Two- Sample T 2 Chater 600 Hotelling s Two- Samle T Introduction This module calculates ower for the Hotelling s two-grou, T-squared (T) test statistic. Hotelling s T is an extension of the univariate two-samle t-test

More information

CHAPTER 5 STATISTICAL INFERENCE. 1.0 Hypothesis Testing. 2.0 Decision Errors. 3.0 How a Hypothesis is Tested. 4.0 Test for Goodness of Fit

CHAPTER 5 STATISTICAL INFERENCE. 1.0 Hypothesis Testing. 2.0 Decision Errors. 3.0 How a Hypothesis is Tested. 4.0 Test for Goodness of Fit Chater 5 Statistical Inference 69 CHAPTER 5 STATISTICAL INFERENCE.0 Hyothesis Testing.0 Decision Errors 3.0 How a Hyothesis is Tested 4.0 Test for Goodness of Fit 5.0 Inferences about Two Means It ain't

More information

Solved Problems. (a) (b) (c) Figure P4.1 Simple Classification Problems First we draw a line between each set of dark and light data points.

Solved Problems. (a) (b) (c) Figure P4.1 Simple Classification Problems First we draw a line between each set of dark and light data points. Solved Problems Solved Problems P Solve the three simle classification roblems shown in Figure P by drawing a decision boundary Find weight and bias values that result in single-neuron ercetrons with the

More information

AI*IA 2003 Fusion of Multiple Pattern Classifiers PART III

AI*IA 2003 Fusion of Multiple Pattern Classifiers PART III AI*IA 23 Fusion of Multile Pattern Classifiers PART III AI*IA 23 Tutorial on Fusion of Multile Pattern Classifiers by F. Roli 49 Methods for fusing multile classifiers Methods for fusing multile classifiers

More information

Research on Evaluation Method of Organization s Performance Based on Comparative Advantage Characteristics

Research on Evaluation Method of Organization s Performance Based on Comparative Advantage Characteristics Vol.1, No.10, Ar 01,.67-7 Research on Evaluation Method of Organization s Performance Based on Comarative Advantage Characteristics WEN Xin 1, JIA Jianfeng and ZHAO Xi nan 3 Abstract It as under the guidance

More information

State Estimation with ARMarkov Models

State Estimation with ARMarkov Models Deartment of Mechanical and Aerosace Engineering Technical Reort No. 3046, October 1998. Princeton University, Princeton, NJ. State Estimation with ARMarkov Models Ryoung K. Lim 1 Columbia University,

More information

A Simple Weight Decay Can Improve. Abstract. It has been observed in numerical simulations that a weight decay can improve

A Simple Weight Decay Can Improve. Abstract. It has been observed in numerical simulations that a weight decay can improve In Advances in Neural Information Processing Systems 4, J.E. Moody, S.J. Hanson and R.P. Limann, eds. Morgan Kaumann Publishers, San Mateo CA, 1995,. 950{957. A Simle Weight Decay Can Imrove Generalization

More information

LIMITATIONS OF RECEPTRON. XOR Problem The failure of the perceptron to successfully simple problem such as XOR (Minsky and Papert).

LIMITATIONS OF RECEPTRON. XOR Problem The failure of the perceptron to successfully simple problem such as XOR (Minsky and Papert). LIMITATIONS OF RECEPTRON XOR Problem The failure of the ercetron to successfully simle roblem such as XOR (Minsky and Paert). x y z x y z 0 0 0 0 0 0 Fig. 4. The exclusive-or logic symbol and function

More information

Use of Transformations and the Repeated Statement in PROC GLM in SAS Ed Stanek

Use of Transformations and the Repeated Statement in PROC GLM in SAS Ed Stanek Use of Transformations and the Reeated Statement in PROC GLM in SAS Ed Stanek Introduction We describe how the Reeated Statement in PROC GLM in SAS transforms the data to rovide tests of hyotheses of interest.

More information

Tests for Two Proportions in a Stratified Design (Cochran/Mantel-Haenszel Test)

Tests for Two Proportions in a Stratified Design (Cochran/Mantel-Haenszel Test) Chater 225 Tests for Two Proortions in a Stratified Design (Cochran/Mantel-Haenszel Test) Introduction In a stratified design, the subects are selected from two or more strata which are formed from imortant

More information

A MIXED CONTROL CHART ADAPTED TO THE TRUNCATED LIFE TEST BASED ON THE WEIBULL DISTRIBUTION

A MIXED CONTROL CHART ADAPTED TO THE TRUNCATED LIFE TEST BASED ON THE WEIBULL DISTRIBUTION O P E R A T I O N S R E S E A R C H A N D D E C I S I O N S No. 27 DOI:.5277/ord73 Nasrullah KHAN Muhammad ASLAM 2 Kyung-Jun KIM 3 Chi-Hyuck JUN 4 A MIXED CONTROL CHART ADAPTED TO THE TRUNCATED LIFE TEST

More information

General Linear Model Introduction, Classes of Linear models and Estimation

General Linear Model Introduction, Classes of Linear models and Estimation Stat 740 General Linear Model Introduction, Classes of Linear models and Estimation An aim of scientific enquiry: To describe or to discover relationshis among events (variables) in the controlled (laboratory)

More information

On Line Parameter Estimation of Electric Systems using the Bacterial Foraging Algorithm

On Line Parameter Estimation of Electric Systems using the Bacterial Foraging Algorithm On Line Parameter Estimation of Electric Systems using the Bacterial Foraging Algorithm Gabriel Noriega, José Restreo, Víctor Guzmán, Maribel Giménez and José Aller Universidad Simón Bolívar Valle de Sartenejas,

More information

Learning Sequence Motif Models Using Gibbs Sampling

Learning Sequence Motif Models Using Gibbs Sampling Learning Sequence Motif Models Using Gibbs Samling BMI/CS 776 www.biostat.wisc.edu/bmi776/ Sring 2018 Anthony Gitter gitter@biostat.wisc.edu These slides excluding third-arty material are licensed under

More information

An Algorithm of Two-Phase Learning for Eleman Neural Network to Avoid the Local Minima Problem

An Algorithm of Two-Phase Learning for Eleman Neural Network to Avoid the Local Minima Problem IJCSNS International Journal of Comuter Science and Networ Security, VOL.7 No.4, Aril 2007 1 An Algorithm of Two-Phase Learning for Eleman Neural Networ to Avoid the Local Minima Problem Zhiqiang Zhang,

More information

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Technical Sciences and Alied Mathematics MODELING THE RELIABILITY OF CISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Cezar VASILESCU Regional Deartment of Defense Resources Management

More information

A Qualitative Event-based Approach to Multiple Fault Diagnosis in Continuous Systems using Structural Model Decomposition

A Qualitative Event-based Approach to Multiple Fault Diagnosis in Continuous Systems using Structural Model Decomposition A Qualitative Event-based Aroach to Multile Fault Diagnosis in Continuous Systems using Structural Model Decomosition Matthew J. Daigle a,,, Anibal Bregon b,, Xenofon Koutsoukos c, Gautam Biswas c, Belarmino

More information

System Reliability Estimation and Confidence Regions from Subsystem and Full System Tests

System Reliability Estimation and Confidence Regions from Subsystem and Full System Tests 009 American Control Conference Hyatt Regency Riverfront, St. Louis, MO, USA June 0-, 009 FrB4. System Reliability Estimation and Confidence Regions from Subsystem and Full System Tests James C. Sall Abstract

More information

Plotting the Wilson distribution

Plotting the Wilson distribution , Survey of English Usage, University College London Setember 018 1 1. Introduction We have discussed the Wilson score interval at length elsewhere (Wallis 013a, b). Given an observed Binomial roortion

More information

Evaluating Circuit Reliability Under Probabilistic Gate-Level Fault Models

Evaluating Circuit Reliability Under Probabilistic Gate-Level Fault Models Evaluating Circuit Reliability Under Probabilistic Gate-Level Fault Models Ketan N. Patel, Igor L. Markov and John P. Hayes University of Michigan, Ann Arbor 48109-2122 {knatel,imarkov,jhayes}@eecs.umich.edu

More information

Recent Developments in Multilayer Perceptron Neural Networks

Recent Developments in Multilayer Perceptron Neural Networks Recent Develoments in Multilayer Percetron eural etworks Walter H. Delashmit Lockheed Martin Missiles and Fire Control Dallas, Texas 75265 walter.delashmit@lmco.com walter.delashmit@verizon.net Michael

More information

Introduction to Probability and Statistics

Introduction to Probability and Statistics Introduction to Probability and Statistics Chater 8 Ammar M. Sarhan, asarhan@mathstat.dal.ca Deartment of Mathematics and Statistics, Dalhousie University Fall Semester 28 Chater 8 Tests of Hyotheses Based

More information

Keywords: pile, liquefaction, lateral spreading, analysis ABSTRACT

Keywords: pile, liquefaction, lateral spreading, analysis ABSTRACT Key arameters in seudo-static analysis of iles in liquefying sand Misko Cubrinovski Deartment of Civil Engineering, University of Canterbury, Christchurch 814, New Zealand Keywords: ile, liquefaction,

More information

A Comparison between Biased and Unbiased Estimators in Ordinary Least Squares Regression

A Comparison between Biased and Unbiased Estimators in Ordinary Least Squares Regression Journal of Modern Alied Statistical Methods Volume Issue Article 7 --03 A Comarison between Biased and Unbiased Estimators in Ordinary Least Squares Regression Ghadban Khalaf King Khalid University, Saudi

More information

Probability Estimates for Multi-class Classification by Pairwise Coupling

Probability Estimates for Multi-class Classification by Pairwise Coupling Probability Estimates for Multi-class Classification by Pairwise Couling Ting-Fan Wu Chih-Jen Lin Deartment of Comuter Science National Taiwan University Taiei 06, Taiwan Ruby C. Weng Deartment of Statistics

More information

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK Towards understanding the Lorenz curve using the Uniform distribution Chris J. Stehens Newcastle City Council, Newcastle uon Tyne, UK (For the Gini-Lorenz Conference, University of Siena, Italy, May 2005)

More information

Principal Components Analysis and Unsupervised Hebbian Learning

Principal Components Analysis and Unsupervised Hebbian Learning Princial Comonents Analysis and Unsuervised Hebbian Learning Robert Jacobs Deartment of Brain & Cognitive Sciences University of Rochester Rochester, NY 1467, USA August 8, 008 Reference: Much of the material

More information

Bayesian Model Averaging Kriging Jize Zhang and Alexandros Taflanidis

Bayesian Model Averaging Kriging Jize Zhang and Alexandros Taflanidis HIPAD LAB: HIGH PERFORMANCE SYSTEMS LABORATORY DEPARTMENT OF CIVIL AND ENVIRONMENTAL ENGINEERING AND EARTH SCIENCES Bayesian Model Averaging Kriging Jize Zhang and Alexandros Taflanidis Why use metamodeling

More information

HEAT, WORK, AND THE FIRST LAW OF THERMODYNAMICS

HEAT, WORK, AND THE FIRST LAW OF THERMODYNAMICS HET, ORK, ND THE FIRST L OF THERMODYNMIS 8 EXERISES Section 8. The First Law of Thermodynamics 5. INTERPRET e identify the system as the water in the insulated container. The roblem involves calculating

More information

An Analysis of Reliable Classifiers through ROC Isometrics

An Analysis of Reliable Classifiers through ROC Isometrics An Analysis of Reliable Classifiers through ROC Isometrics Stijn Vanderlooy s.vanderlooy@cs.unimaas.nl Ida G. Srinkhuizen-Kuyer kuyer@cs.unimaas.nl Evgueni N. Smirnov smirnov@cs.unimaas.nl MICC-IKAT, Universiteit

More information

MULTIVARIATE STATISTICAL PROCESS OF HOTELLING S T CONTROL CHARTS PROCEDURES WITH INDUSTRIAL APPLICATION

MULTIVARIATE STATISTICAL PROCESS OF HOTELLING S T CONTROL CHARTS PROCEDURES WITH INDUSTRIAL APPLICATION Journal of Statistics: Advances in heory and Alications Volume 8, Number, 07, Pages -44 Available at htt://scientificadvances.co.in DOI: htt://dx.doi.org/0.864/jsata_700868 MULIVARIAE SAISICAL PROCESS

More information

GENERATING FUZZY RULES FOR PROTEIN CLASSIFICATION E. G. MANSOORI, M. J. ZOLGHADRI, S. D. KATEBI, H. MOHABATKAR, R. BOOSTANI AND M. H.

GENERATING FUZZY RULES FOR PROTEIN CLASSIFICATION E. G. MANSOORI, M. J. ZOLGHADRI, S. D. KATEBI, H. MOHABATKAR, R. BOOSTANI AND M. H. Iranian Journal of Fuzzy Systems Vol. 5, No. 2, (2008). 21-33 GENERATING FUZZY RULES FOR PROTEIN CLASSIFICATION E. G. MANSOORI, M. J. ZOLGHADRI, S. D. KATEBI, H. MOHABATKAR, R. BOOSTANI AND M. H. SADREDDINI

More information

arxiv:cond-mat/ v2 25 Sep 2002

arxiv:cond-mat/ v2 25 Sep 2002 Energy fluctuations at the multicritical oint in two-dimensional sin glasses arxiv:cond-mat/0207694 v2 25 Se 2002 1. Introduction Hidetoshi Nishimori, Cyril Falvo and Yukiyasu Ozeki Deartment of Physics,

More information

Chapter 1 Fundamentals

Chapter 1 Fundamentals Chater Fundamentals. Overview of Thermodynamics Industrial Revolution brought in large scale automation of many tedious tasks which were earlier being erformed through manual or animal labour. Inventors

More information

University of North Carolina-Charlotte Department of Electrical and Computer Engineering ECGR 4143/5195 Electrical Machinery Fall 2009

University of North Carolina-Charlotte Department of Electrical and Computer Engineering ECGR 4143/5195 Electrical Machinery Fall 2009 University of North Carolina-Charlotte Deartment of Electrical and Comuter Engineering ECG 4143/5195 Electrical Machinery Fall 9 Problem Set 5 Part Due: Friday October 3 Problem 3: Modeling the exerimental

More information

RUN-TO-RUN CONTROL AND PERFORMANCE MONITORING OF OVERLAY IN SEMICONDUCTOR MANUFACTURING. 3 Department of Chemical Engineering

RUN-TO-RUN CONTROL AND PERFORMANCE MONITORING OF OVERLAY IN SEMICONDUCTOR MANUFACTURING. 3 Department of Chemical Engineering Coyright 2002 IFAC 15th Triennial World Congress, Barcelona, Sain RUN-TO-RUN CONTROL AND PERFORMANCE MONITORING OF OVERLAY IN SEMICONDUCTOR MANUFACTURING C.A. Bode 1, B.S. Ko 2, and T.F. Edgar 3 1 Advanced

More information

8 STOCHASTIC PROCESSES

8 STOCHASTIC PROCESSES 8 STOCHASTIC PROCESSES The word stochastic is derived from the Greek στoχαστικoς, meaning to aim at a target. Stochastic rocesses involve state which changes in a random way. A Markov rocess is a articular

More information

PCA fused NN approach for drill wear prediction in drilling mild steel specimen

PCA fused NN approach for drill wear prediction in drilling mild steel specimen PCA fused NN aroach for drill wear rediction in drilling mild steel secimen S.S. Panda Deartment of Mechanical Engineering, Indian Institute of Technology Patna, Bihar-8000, India, Tel No.: 9-99056477(M);

More information

NUMERICAL AND THEORETICAL INVESTIGATIONS ON DETONATION- INERT CONFINEMENT INTERACTIONS

NUMERICAL AND THEORETICAL INVESTIGATIONS ON DETONATION- INERT CONFINEMENT INTERACTIONS NUMERICAL AND THEORETICAL INVESTIGATIONS ON DETONATION- INERT CONFINEMENT INTERACTIONS Tariq D. Aslam and John B. Bdzil Los Alamos National Laboratory Los Alamos, NM 87545 hone: 1-55-667-1367, fax: 1-55-667-6372

More information

Optimizing the Polynomial to Represent the Extended True Boiling Point Curve from High Vacuum Distillation Data Using Genetic Algorithms

Optimizing the Polynomial to Represent the Extended True Boiling Point Curve from High Vacuum Distillation Data Using Genetic Algorithms 1561 A ublication of CHEMICAL ENGINEERING TRANSACTIONS VOL. 43, 015 Chief Editors: Sauro Pierucci, Jiří J. Klemeš Coyright 015, AIDIC Servizi S.r.l., ISBN 978-88-95608-34-1; ISSN 83-916 The Italian Association

More information

The Poisson Regression Model

The Poisson Regression Model The Poisson Regression Model The Poisson regression model aims at modeling a counting variable Y, counting the number of times that a certain event occurs during a given time eriod. We observe a samle

More information

Topology Optimization of Three Dimensional Structures under Self-weight and Inertial Forces

Topology Optimization of Three Dimensional Structures under Self-weight and Inertial Forces 6 th World Congresses of Structural and Multidiscilinary Otimization Rio de Janeiro, 30 May - 03 June 2005, Brazil Toology Otimization of Three Dimensional Structures under Self-weight and Inertial Forces

More information

PHYSICAL REVIEW LETTERS

PHYSICAL REVIEW LETTERS PHYSICAL REVIEW LETTERS VOLUME 81 20 JULY 1998 NUMBER 3 Searated-Path Ramsey Atom Interferometer P. D. Featonby, G. S. Summy, C. L. Webb, R. M. Godun, M. K. Oberthaler, A. C. Wilson, C. J. Foot, and K.

More information

Uniformly best wavenumber approximations by spatial central difference operators: An initial investigation

Uniformly best wavenumber approximations by spatial central difference operators: An initial investigation Uniformly best wavenumber aroximations by satial central difference oerators: An initial investigation Vitor Linders and Jan Nordström Abstract A characterisation theorem for best uniform wavenumber aroximations

More information

The Noise Power Ratio - Theory and ADC Testing

The Noise Power Ratio - Theory and ADC Testing The Noise Power Ratio - Theory and ADC Testing FH Irons, KJ Riley, and DM Hummels Abstract This aer develos theory behind the noise ower ratio (NPR) testing of ADCs. A mid-riser formulation is used for

More information

KEY ISSUES IN THE ANALYSIS OF PILES IN LIQUEFYING SOILS

KEY ISSUES IN THE ANALYSIS OF PILES IN LIQUEFYING SOILS 4 th International Conference on Earthquake Geotechnical Engineering June 2-28, 27 KEY ISSUES IN THE ANALYSIS OF PILES IN LIQUEFYING SOILS Misko CUBRINOVSKI 1, Hayden BOWEN 1 ABSTRACT Two methods for analysis

More information

Machine Learning: Homework 4

Machine Learning: Homework 4 10-601 Machine Learning: Homework 4 Due 5.m. Monday, February 16, 2015 Instructions Late homework olicy: Homework is worth full credit if submitted before the due date, half credit during the next 48 hours,

More information

Modeling and Estimation of Full-Chip Leakage Current Considering Within-Die Correlation

Modeling and Estimation of Full-Chip Leakage Current Considering Within-Die Correlation 6.3 Modeling and Estimation of Full-Chi Leaage Current Considering Within-Die Correlation Khaled R. eloue, Navid Azizi, Farid N. Najm Deartment of ECE, University of Toronto,Toronto, Ontario, Canada {haled,nazizi,najm}@eecg.utoronto.ca

More information

MODEL-BASED MULTIPLE FAULT DETECTION AND ISOLATION FOR NONLINEAR SYSTEMS

MODEL-BASED MULTIPLE FAULT DETECTION AND ISOLATION FOR NONLINEAR SYSTEMS MODEL-BASED MULIPLE FAUL DEECION AND ISOLAION FOR NONLINEAR SYSEMS Ivan Castillo, and homas F. Edgar he University of exas at Austin Austin, X 78712 David Hill Chemstations Houston, X 77009 Abstract A

More information

A New GP-evolved Formulation for the Relative Permittivity of Water and Steam

A New GP-evolved Formulation for the Relative Permittivity of Water and Steam ew GP-evolved Formulation for the Relative Permittivity of Water and Steam S. V. Fogelson and W. D. Potter rtificial Intelligence Center he University of Georgia, US Contact Email ddress: sergeyf1@uga.edu

More information

Phase velocity and group velocity (c) Zhengqing Yun,

Phase velocity and group velocity (c) Zhengqing Yun, Phase velocity and grou velocity (c) Zhengqing Yun, 2011-2012 Objective: Observe the difference between hase and grou velocity; understand that the grou velocity can be less than, equal to, and greater

More information

Participation Factors. However, it does not give the influence of each state on the mode.

Participation Factors. However, it does not give the influence of each state on the mode. Particiation Factors he mode shae, as indicated by the right eigenvector, gives the relative hase of each state in a articular mode. However, it does not give the influence of each state on the mode. We

More information

Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition

Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition Haiing Lu, K.N. Plataniotis and A.N. Venetsanooulos The Edward S. Rogers Sr. Deartment of

More information

Pretest (Optional) Use as an additional pacing tool to guide instruction. August 21

Pretest (Optional) Use as an additional pacing tool to guide instruction. August 21 Trimester 1 Pretest (Otional) Use as an additional acing tool to guide instruction. August 21 Beyond the Basic Facts In Trimester 1, Grade 8 focus on multilication. Daily Unit 1: Rational vs. Irrational

More information

Using a Computational Intelligence Hybrid Approach to Recognize the Faults of Variance Shifts for a Manufacturing Process

Using a Computational Intelligence Hybrid Approach to Recognize the Faults of Variance Shifts for a Manufacturing Process Journal of Industrial and Intelligent Information Vol. 4, No. 2, March 26 Using a Comutational Intelligence Hybrid Aroach to Recognize the Faults of Variance hifts for a Manufacturing Process Yuehjen E.

More information

Study on Characteristics of Sound Absorption of Underwater Visco-elastic Coated Compound Structures

Study on Characteristics of Sound Absorption of Underwater Visco-elastic Coated Compound Structures Vol. 3, No. Modern Alied Science Study on Characteristics of Sound Absortion of Underwater Visco-elastic Coated Comound Structures Zhihong Liu & Meiing Sheng College of Marine Northwestern Polytechnical

More information

Uncorrelated Multilinear Principal Component Analysis for Unsupervised Multilinear Subspace Learning

Uncorrelated Multilinear Principal Component Analysis for Unsupervised Multilinear Subspace Learning TNN-2009-P-1186.R2 1 Uncorrelated Multilinear Princial Comonent Analysis for Unsuervised Multilinear Subsace Learning Haiing Lu, K. N. Plataniotis and A. N. Venetsanooulos The Edward S. Rogers Sr. Deartment

More information

Hidden Predictors: A Factor Analysis Primer

Hidden Predictors: A Factor Analysis Primer Hidden Predictors: A Factor Analysis Primer Ryan C Sanchez Western Washington University Factor Analysis is a owerful statistical method in the modern research sychologist s toolbag When used roerly, factor

More information

Combining Logistic Regression with Kriging for Mapping the Risk of Occurrence of Unexploded Ordnance (UXO)

Combining Logistic Regression with Kriging for Mapping the Risk of Occurrence of Unexploded Ordnance (UXO) Combining Logistic Regression with Kriging for Maing the Risk of Occurrence of Unexloded Ordnance (UXO) H. Saito (), P. Goovaerts (), S. A. McKenna (2) Environmental and Water Resources Engineering, Deartment

More information

Improved Capacity Bounds for the Binary Energy Harvesting Channel

Improved Capacity Bounds for the Binary Energy Harvesting Channel Imroved Caacity Bounds for the Binary Energy Harvesting Channel Kaya Tutuncuoglu 1, Omur Ozel 2, Aylin Yener 1, and Sennur Ulukus 2 1 Deartment of Electrical Engineering, The Pennsylvania State University,

More information

1-way quantum finite automata: strengths, weaknesses and generalizations

1-way quantum finite automata: strengths, weaknesses and generalizations 1-way quantum finite automata: strengths, weaknesses and generalizations arxiv:quant-h/9802062v3 30 Se 1998 Andris Ambainis UC Berkeley Abstract Rūsiņš Freivalds University of Latvia We study 1-way quantum

More information

A Game Theoretic Investigation of Selection Methods in Two Population Coevolution

A Game Theoretic Investigation of Selection Methods in Two Population Coevolution A Game Theoretic Investigation of Selection Methods in Two Poulation Coevolution Sevan G. Ficici Division of Engineering and Alied Sciences Harvard University Cambridge, Massachusetts 238 USA sevan@eecs.harvard.edu

More information

RANDOM WALKS AND PERCOLATION: AN ANALYSIS OF CURRENT RESEARCH ON MODELING NATURAL PROCESSES

RANDOM WALKS AND PERCOLATION: AN ANALYSIS OF CURRENT RESEARCH ON MODELING NATURAL PROCESSES RANDOM WALKS AND PERCOLATION: AN ANALYSIS OF CURRENT RESEARCH ON MODELING NATURAL PROCESSES AARON ZWIEBACH Abstract. In this aer we will analyze research that has been recently done in the field of discrete

More information

Sensitivity and Robustness of Quantum Spin-½ Rings to Parameter Uncertainty

Sensitivity and Robustness of Quantum Spin-½ Rings to Parameter Uncertainty Sensitivity and Robustness of Quantum Sin-½ Rings to Parameter Uncertainty Sean O Neil, Member, IEEE, Edmond Jonckheere, Life Fellow, IEEE, Sohie Schirmer, Member, IEEE, and Frank Langbein, Member, IEEE

More information

The Numerical Simulation of Gas Turbine Inlet-Volute Flow Field

The Numerical Simulation of Gas Turbine Inlet-Volute Flow Field World Journal of Mechanics, 013, 3, 30-35 doi:10.436/wjm.013.3403 Published Online July 013 (htt://www.scir.org/journal/wjm) The Numerical Simulation of Gas Turbine Inlet-Volute Flow Field Tao Jiang 1,

More information

Scaling Multiple Point Statistics for Non-Stationary Geostatistical Modeling

Scaling Multiple Point Statistics for Non-Stationary Geostatistical Modeling Scaling Multile Point Statistics or Non-Stationary Geostatistical Modeling Julián M. Ortiz, Steven Lyster and Clayton V. Deutsch Centre or Comutational Geostatistics Deartment o Civil & Environmental Engineering

More information

Roots Blower with Gradually Expanding Outlet Gap: Mathematical Modelling and Performance Simulation Yingjie Cai 1, a, Ligang Yao 2, b

Roots Blower with Gradually Expanding Outlet Gap: Mathematical Modelling and Performance Simulation Yingjie Cai 1, a, Ligang Yao 2, b 2nd International Conference on Advances in Mechanical Engineering and Industrial Informatics (AMEII 216) Roots Bloer ith Gradually Exanding Outlet Ga: Mathematical Modelling and Performance Simulation

More information

Chapter 7 Sampling and Sampling Distributions. Introduction. Selecting a Sample. Introduction. Sampling from a Finite Population

Chapter 7 Sampling and Sampling Distributions. Introduction. Selecting a Sample. Introduction. Sampling from a Finite Population Chater 7 and s Selecting a Samle Point Estimation Introduction to s of Proerties of Point Estimators Other Methods Introduction An element is the entity on which data are collected. A oulation is a collection

More information

arxiv: v2 [physics.soc-ph] 5 Mar 2012

arxiv: v2 [physics.soc-ph] 5 Mar 2012 Robustness of onion-like correlated networks against targeted attacks Toshihiro Tanizawa, 1, hlomo Havlin, 2 and H. Eugene tanley 3 1 Kochi National College of Technology, 200-1 Monobe-Otsu, Nankoku, Kochi

More information

Named Entity Recognition using Maximum Entropy Model SEEM5680

Named Entity Recognition using Maximum Entropy Model SEEM5680 Named Entity Recognition using Maximum Entroy Model SEEM5680 Named Entity Recognition System Named Entity Recognition (NER): Identifying certain hrases/word sequences in a free text. Generally it involves

More information

Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition

Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition TNN-2007-P-0332.R1 1 Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition Haiing Lu, K.N. Plataniotis and A.N. Venetsanooulos The Edward S. Rogers

More information

One step ahead prediction using Fuzzy Boolean Neural Networks 1

One step ahead prediction using Fuzzy Boolean Neural Networks 1 One ste ahead rediction using Fuzzy Boolean eural etworks 1 José A. B. Tomé IESC-ID, IST Rua Alves Redol, 9 1000 Lisboa jose.tome@inesc-id.t João Paulo Carvalho IESC-ID, IST Rua Alves Redol, 9 1000 Lisboa

More information

Session 5: Review of Classical Astrodynamics

Session 5: Review of Classical Astrodynamics Session 5: Review of Classical Astrodynamics In revious lectures we described in detail the rocess to find the otimal secific imulse for a articular situation. Among the mission requirements that serve

More information

ON SETS OF INTEGERS WHICH CONTAIN NO THREE TERMS IN GEOMETRIC PROGRESSION

ON SETS OF INTEGERS WHICH CONTAIN NO THREE TERMS IN GEOMETRIC PROGRESSION ON SETS OF INTEGERS WHICH CONTAIN NO THREE TERMS IN GEOMETRIC PROGRESSION NATHAN MCNEW Abstract The roblem of looing for subsets of the natural numbers which contain no 3-term arithmetic rogressions has

More information

Deriving Indicator Direct and Cross Variograms from a Normal Scores Variogram Model (bigaus-full) David F. Machuca Mory and Clayton V.

Deriving Indicator Direct and Cross Variograms from a Normal Scores Variogram Model (bigaus-full) David F. Machuca Mory and Clayton V. Deriving ndicator Direct and Cross Variograms from a Normal Scores Variogram Model (bigaus-full) David F. Machuca Mory and Clayton V. Deutsch Centre for Comutational Geostatistics Deartment of Civil &

More information

Uncertainty Modeling with Interval Type-2 Fuzzy Logic Systems in Mobile Robotics

Uncertainty Modeling with Interval Type-2 Fuzzy Logic Systems in Mobile Robotics Uncertainty Modeling with Interval Tye-2 Fuzzy Logic Systems in Mobile Robotics Ondrej Linda, Student Member, IEEE, Milos Manic, Senior Member, IEEE bstract Interval Tye-2 Fuzzy Logic Systems (IT2 FLSs)

More information

Using the Divergence Information Criterion for the Determination of the Order of an Autoregressive Process

Using the Divergence Information Criterion for the Determination of the Order of an Autoregressive Process Using the Divergence Information Criterion for the Determination of the Order of an Autoregressive Process P. Mantalos a1, K. Mattheou b, A. Karagrigoriou b a.deartment of Statistics University of Lund

More information

pp physics, RWTH, WS 2003/04, T.Hebbeker

pp physics, RWTH, WS 2003/04, T.Hebbeker 1. PP TH 03/04 Accelerators and Detectors 1 hysics, RWTH, WS 2003/04, T.Hebbeker 2003-12-03 1. Accelerators and Detectors In the following, we concentrate on the three machines SPS, Tevatron and LHC with

More information

Time Domain Calculation of Vortex Induced Vibration of Long-Span Bridges by Using a Reduced-order Modeling Technique

Time Domain Calculation of Vortex Induced Vibration of Long-Span Bridges by Using a Reduced-order Modeling Technique 2017 2nd International Conference on Industrial Aerodynamics (ICIA 2017) ISBN: 978-1-60595-481-3 Time Domain Calculation of Vortex Induced Vibration of Long-San Bridges by Using a Reduced-order Modeling

More information

Seafloor Reflectivity A Test of an Inversion Technique

Seafloor Reflectivity A Test of an Inversion Technique Seafloor Reflectivity A Test of an Inversion Technique Adrian D. Jones 1, Justin Hoffman and Paul A. Clarke 1 1 Defence Science and Technology Organisation, Australia, Student at Centre for Marine Science

More information

Characteristics of Fibonacci-type Sequences

Characteristics of Fibonacci-type Sequences Characteristics of Fibonacci-tye Sequences Yarden Blausa May 018 Abstract This aer resents an exloration of the Fibonacci sequence, as well as multi-nacci sequences and the Lucas sequence. We comare and

More information

Robust Predictive Control of Input Constraints and Interference Suppression for Semi-Trailer System

Robust Predictive Control of Input Constraints and Interference Suppression for Semi-Trailer System Vol.7, No.7 (4),.37-38 htt://dx.doi.org/.457/ica.4.7.7.3 Robust Predictive Control of Inut Constraints and Interference Suression for Semi-Trailer System Zhao, Yang Electronic and Information Technology

More information

Journal of Chemical and Pharmaceutical Research, 2014, 6(5): Research Article

Journal of Chemical and Pharmaceutical Research, 2014, 6(5): Research Article Available online www.jocr.com Journal of Chemical and Pharmaceutical Research, 204, 6(5):580-585 Research Article ISSN : 0975-7384 CODEN(USA) : JCPRC5 Exort facilitation and comarative advantages of the

More information

A SIMPLE PLASTICITY MODEL FOR PREDICTING TRANSVERSE COMPOSITE RESPONSE AND FAILURE

A SIMPLE PLASTICITY MODEL FOR PREDICTING TRANSVERSE COMPOSITE RESPONSE AND FAILURE THE 19 TH INTERNATIONAL CONFERENCE ON COMPOSITE MATERIALS A SIMPLE PLASTICITY MODEL FOR PREDICTING TRANSVERSE COMPOSITE RESPONSE AND FAILURE K.W. Gan*, M.R. Wisnom, S.R. Hallett, G. Allegri Advanced Comosites

More information

602 ZHANG Zhi and CHEN Li-Rong Vol Gibbs Free Energy From the thermodynamics, the Gibbs free energy of the system can be written as G = U ; TS +

602 ZHANG Zhi and CHEN Li-Rong Vol Gibbs Free Energy From the thermodynamics, the Gibbs free energy of the system can be written as G = U ; TS + Commun. Theor. Phys. (Beijing, China) 37 (2002) 601{606 c International Academic Publishers Vol. 37, No. 5, May 15, 2002 The 2D Alternative Binary L J System: Solid-Liquid Phase Diagram ZHANG Zhi and CHEN

More information

Optical Fibres - Dispersion Part 1

Optical Fibres - Dispersion Part 1 ECE 455 Lecture 05 1 Otical Fibres - Disersion Part 1 Stavros Iezekiel Deartment of Electrical and Comuter Engineering University of Cyrus HMY 445 Lecture 05 Fall Semester 016 ECE 455 Lecture 05 Otical

More information

PRELIMINARY SELECTION OF BASIC PARAMETERS OF DIFFERENT TORSIONAL VIBRATION DAMPERS INTENDED FOR USE IN MEDIUM-SPEED DIESEL ENGINES

PRELIMINARY SELECTION OF BASIC PARAMETERS OF DIFFERENT TORSIONAL VIBRATION DAMPERS INTENDED FOR USE IN MEDIUM-SPEED DIESEL ENGINES Ivan Filiović Dževad Bibić Aleksandar Milašinović Almir Blažević Adnan Pecar ISSN 333-4 PRELIMINARY SELECTION OF BASIC PARAMETERS OF DIFFERENT TORSIONAL VIBRATION DAMPERS INTENDED FOR USE IN MEDIUM-SPEED

More information

Topic: Lower Bounds on Randomized Algorithms Date: September 22, 2004 Scribe: Srinath Sridhar

Topic: Lower Bounds on Randomized Algorithms Date: September 22, 2004 Scribe: Srinath Sridhar 15-859(M): Randomized Algorithms Lecturer: Anuam Guta Toic: Lower Bounds on Randomized Algorithms Date: Setember 22, 2004 Scribe: Srinath Sridhar 4.1 Introduction In this lecture, we will first consider

More information

COMMUNICATION BETWEEN SHAREHOLDERS 1

COMMUNICATION BETWEEN SHAREHOLDERS 1 COMMUNICATION BTWN SHARHOLDRS 1 A B. O A : A D Lemma B.1. U to µ Z r 2 σ2 Z + σ2 X 2r ω 2 an additive constant that does not deend on a or θ, the agents ayoffs can be written as: 2r rθa ω2 + θ µ Y rcov

More information

STABILITY ANALYSIS AND CONTROL OF STOCHASTIC DYNAMIC SYSTEMS USING POLYNOMIAL CHAOS. A Dissertation JAMES ROBERT FISHER

STABILITY ANALYSIS AND CONTROL OF STOCHASTIC DYNAMIC SYSTEMS USING POLYNOMIAL CHAOS. A Dissertation JAMES ROBERT FISHER STABILITY ANALYSIS AND CONTROL OF STOCHASTIC DYNAMIC SYSTEMS USING POLYNOMIAL CHAOS A Dissertation by JAMES ROBERT FISHER Submitted to the Office of Graduate Studies of Texas A&M University in artial fulfillment

More information

Multivariable Generalized Predictive Scheme for Gas Turbine Control in Combined Cycle Power Plant

Multivariable Generalized Predictive Scheme for Gas Turbine Control in Combined Cycle Power Plant Multivariable Generalized Predictive Scheme for Gas urbine Control in Combined Cycle Power Plant L.X.Niu and X.J.Liu Deartment of Automation North China Electric Power University Beiing, China, 006 e-mail

More information