5:25 PM Sun Oct 23"DoneWSEVANEMATE VAN54.↑Y-axis10095908580757065605550454035302520 A1510SVM decision boundary. Consider again an SVM whose decision boundary is obtained byEquations 1 and 2 like in the previous questions. Now, consider the training dataset with two trainingclasses (signaled as squares and triangles) given by the scatter plot in the following figure. Draw thedecision boundary for this dataset obtained by our SVM on it. The drawing needs to correctly separatethe examples but there are multiple possible solutions. Each point of the line you draw must matchthe points of the true decision boundary within the error margin of 10 unit intervals along the x-axisand y-axis.500-480AA●●●hw3_handout50AA9A.….……………….AIU00 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100X-axis69+010:@ 97% Consider an SVM that obtains a decision boundary through the maximization optimization in Equation (1),which is equivalent to the minimization optimization in Equation (2), where 3 and 3o denote the parameters,x the i-th training example, y, € {−1, 1} the class label for the i-th training example, M the margin, N thenumber of instances.which is equivalent to1.M2.maxB,Bo,||B||2=1subject to y (x + ßo) ≥ M, i = 1, ..., N.min=||||B,Bo 2subject to y (x + ßo) ≥ 1, i = 1,..., N.Mark true or false for the following assertions (Q4.1, Q4.2, Q4.3) and justify your answer. Questions withno justification will receive 0 points.(1)This SVM objective assumes data is linearly separable.(2)After obtaining the decision boundary in Equation 1, if we modify ß while maintaining therestriction ||3|| = 1, this would modify the minimum distance from the decision boundary to theorigin.3.= 2Assume the maximum margin obtained by the model was M = 4 = d₁ + d2, where d₁and d₂ = 2 denote the distance from the decision boundary to hyperplanes H₁ and H₂ defined by thesupport vectors. If we add 1 to Bo, the new maximum margin would be M + 1.

4. 9 SVM decision boundary. Consider again an SVM whose decision boundary is obtained by Equations 1 and 2 like in the previous questions. Now, consider the training dataset with two training classes (signaled as squares and triangles) given by the scatter plot in the following figure. Draw the decision boundary for this dataset obtained by our SVM on it. The drawing needs to correctly separate the examples but there are multiple possible solutions. Each point of the line you draw must match the points of the true decision boundary within the error margin of 10 unit intervals along the x-axis and y-axis. 100 95 90 85 80 75 70 65 60 55 A 0 0 P

4. 9 SVM decision boundary. Consider again an SVM whose decision boundary is obtained by Equations 1 and 2 like in the previous questions. Now, consider the training dataset with two training classes (signaled as squares and triangles) given by the scatter plot in the following figure. Draw the decision boundary for this dataset obtained by our SVM on it. The drawing needs to correctly separate the examples but there are multiple possible solutions. Each point of the line you draw must match the points of the true decision boundary within the error margin of 10 unit intervals along the x-axis and y-axis. 100 95 90 85 80 75 70 65 60 55 A 0 0 P

Database System Concepts

7th Edition

ISBN:9780078022159

Author:Abraham Silberschatz Professor, Henry F. Korth, S. Sudarshan

Publisher:Abraham Silberschatz Professor, Henry F. Korth, S. Sudarshan

Chapter1: Introduction

Section: Chapter Questions

Problem 1PE

See similar textbooks

Similar questions

Problem 6: Suppose, you are given with a dataset of 5,000 images. Split the dataset into Training and Validation set using 6-fuld cross validation. Hint: How many images will be in training set and how many in validation set.
Consider a collection C with 100,000 documents. “Riverside” and “California” have 500 and 1000 occurrences in C. Each document in C has on average 1000 words. One of these documents is D1: University of California, Riverside, or UC Riverside Consider query Q: Riverside California 1. What is the score of D1 for Q if we use unigram Language Model-based ranking and Jeliker-Mercer Smoothing with lambda=0.5?
igure P2.3 shows a two-dimensional set of data points. Part of the data points belongs to class 61 and the other part belongs to class C,. Construct the decision boundary pro- auced by the nearest neighbor rule applied to this data sample. x :class 6, • : class 62 X2 X1 FIGURE P2.3
Consider a piece of text in which the letters a,e,g,k,l,z occur with probabilities of 3, 8, 13, 19,23,34 percent. Generate a Huffman table to code them.
The non-parametric density-based approach assumes that the density around a normal data observation within a cluster (relatively big) is similar to the density around its neighbours, and the density around an outlier (relatively small) is considerably different to the density around its neighbours. If we had the density of observation within a cluster smaller than the density around an outlier, explain why we would have such a situation. And provide a solution to this problem.
Question 48. Let us return to the Titanic data set. We now have learned several models and want to choose the best one. We used three different methods to validate these models: The training error rate (apparent error rate), the error rate on an external test set and the error rate estimated by a 10-fold cross validation. Training Error | Error on the test set | Cross Validation Error 0.18 Learner Decision Tree 0.22 0.21 Random Forest 0.01 0.10 0.12 1-Nearest-Neighbour 0.18 0.19 Which of the following statements are correct? a) 1-Nearest-Neighbour has a perfect training error and hence it should be used here. b) Random Forests outperforms both 1-Nearest-Neighbour and the Decision Tree in terms of prediction error. c) Not just in this case, but in general, Cross Validation is the better validation strategy and should always be preferred over the error on a single test set. d) Not just in this case, but in general, Decision Trees always perform worse than Random Forests.
Consider the training dataset given in Table P3.5. a. Construct a decision tree from the given data using "information gain" based splitting. Table P3.5 D1 D2 D3 D4 D5 D6 D7 D8 D9 Day D10 D11 D12 D13 D14 sunny sunny overcast rain rain Outlook rain overcast sunny sunny rain sunny overcast overcast rain hot hot hot mild cool cool cool mild cool mild mild mild hot mild Temperature high high high high normal normal Humidity normal high normal normal normal high normal high (Outlook sunny, Temperature cool, Humidity high, Wind - strong) ▪ weak Wind strong weak weak weak strong strong weak weak weak strong strong weak strong no no yes yes yes no yes no yes yes yes yes yes no PlayTennis
Consider a machine learning problem for which the trainings data (x, y) are {(1, 4.8), (3, 11.3), (5, 17.2)). Using univariate linear regression, what is the corresponding hypothesis hex)? What is the predicted output for an attribute of x = 4? Show all the steps in the calculations.
2. Can you design a binary classification experiment with 100 total population (TP+TN+FP+ FN), with precision (TP/(TP+FP)) of 1/2, with sensitivity (TP/(TP+FN)) of 2/3, and specificity (TN/(FP+TN)) of 3/5? (Please consider the population to consist of 100 individuals.)
In a linear classifier, there can be multiple linear models that can perfectly classify the training data points. SVM would prefer a model that has maximum margins from the data points between both classes. Select one: A. True B. False
Like the Hough transform, RANSAC can be used to identify the best fitting line to a group of points. Given a set of 12 points, how many points are used to create a line model? How is a line model scored to determine whether it is a good fit? What makes RANSAC more robust to outliers?
Draw a QQ (quantile-quantile) plot for the built-in data set, islands, to assess the normality of the observations. Is the data set well-modeled by a normal distribution?