Scribd
Upload
45 views2 pages

Midterm Final Version

Uploaded by

viany lepro
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
45 views2 pages

Midterm Final Version

Uploaded by

viany lepro
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 2
‘Question rpz] 3 4 _ [Terr ‘Grade ISTANBUL AYDIN UNIVERSITY ‘Name-Sumame ‘Shudent ID: Signatore: Course Name: SENS20 MACHINE LEARNING — | Date’Tisme: 11/0472023 ‘Duration: 60 min, Group, Midterm Tecuers THe, FullNane Dv: Givkem SERDES ‘NOTE: Please illin your answers by wing your handwriting on a separate sheet of paper. Take the photos of your solutions, ad them toa new Microsofi Word document, save them asa single“ pf file and upload to ihe system, Or vou can diseetly scan vour solutions and send them as. single “pi0” Bile, (Qip) OL Please answer TrueFalse in the following question. Explain your reasoning in 1 sentences {7 pts] a) Suppose you have a dataset which has 900 cancer-free images and 100 images taken from cancer Faint, IT tn classifier which achieves 8596 accuracy on this date, ca Tsay that ny classifier certain is a food classifier? [71}) For a Logistic Regression leamer, you run gradient descent for 15 iterations with an a = 0.2 and compute the cost function J(0) after each iteration. Ifyou find thatthe cost function J(@) decreases slowly and is still decreasing after 15 iterations, you should decrease the learning rate « fora faster learning process [7]) On a specifi dataset, a project team reports a very low training error nnd they cls that their method is perfectly deseribing hidden pattem in ther data, (2p) Q2. Suppose you are given the following logistic regression classification task: predict the target Y © {0, 1} siven two real valued features Xi € R and X: € IR. After some training, you lear the following decision rule: Predict Lif wy + W.X; +w2X; 20 and Y= 0 otherwise + Plot the decision boundary and label the region where we would predict P= 1 and Y= + Suppose that we Ieamed the above weights using logistic prediction for P( 5 pos] ession. Using this model, what would be our 1, [ Ny, X2)s ive the formulation please [10 pts} (0p) Q3. Consider the following training data (x, y pairs) given at right pat: Suppose the data comes from a model y = cx + noise, for unknown constants ¢ and f . x Use least squares linear regression to find an estimate of ¢ and 1/3 Hint: You Can use logarithm to convert y to linear relation. 2/4 3/05 4p) 4. ‘You want to build a model to predict whether a startup will raise funding (label = 1) or not (label = 0) in its first year. You have access to a dataset of examples with 300 input features, including the number of founders, the number of employees, the variance in the age of the employees, ete. As a baseline model, you decide to train a logistic regression that outputs a probability 90 ¢ [0,1] that a startup described by a vector of features 2 raises funding in its first year, where the predicted label of an input is chosen to be 1 when 9 > 0.5 and 0 otherwis a If the shape of x" is (n,,1), what should n, be? » A logistic regression has trainable weights w and bias b. How many weights does the logistic regression model have? 9 What is the dimension of 6? % Consider the prediction 9 € (0,1). ‘What is the dimension of j!? Write the output 9 in terms of the input 2“ and the parameters w and b.

You might also like