Assignment 1 Course Title: Machine Learning Course Code: MTC103 Faculty In charge: Ms Shilpi Gupta Submission Deadline: 02-09-13 Total Marks: 5 • This is a group assignment. Create a group of 2/3 students • Late submissions will NOT be accepted. • Give comprehensive solution for each question. • It is important to you write the answers to all the questions in your own words • Assignments that have been copied and shared among students will be rejected. • All the assignment must be submitted to CR of Class in soft copy. • Assignment submissions are accepted only in .doc or .docx format. Save the file with your names. • Take the assignment seriously; the marks earned in this assignment will be credited in final evaluation. 1) How can we overcome the limitations of Bayesian Belief Networks? Define yourself a real life scenario (apart from the problems discussed in class) which should have at least 5 Nodes. Solve the problem using Belief nets. 1 2) Choose a problem that interests you. Create a data set (at least 10 instances) either your own or you can take online available data set. Select features that you will use to represent the data of your problem? Even if you are using an existing data set, you might select only the features that are relevant to your problem. Solve the problem using K Nearest Neighbor. 2 3) Choose a problem that interests you. Create a data set (at least 10 instances) either your own or you can take online available data set. Select features that you will use to represent the data of your problem? Even if you are using an existing data set, you might select only the features that are relevant to your problem. Solve the problem using Naïve byes Classifier. 2 (Self Study Assignment) 4) What happens when the target we want to learn is noisy? 5) What are the choices available for error measure in machine learning? 6) What is the theory of generalization in machine learning? 7) What are 3 major pitfalls by machine learning practitioners? Hint : 3 major pitfalls by machine learning practitioners Occam’s razor Data snooping Sampling Bias 8) Discuss the bias Variance tradeoff? Why this is considered as an important issue in Machine learning theory? 9) Discuss the following terms– • Over fitting • Under fitting • False Negative • False Positive • Predictive accuracy 10) What is No free lunch rule in machine learning? Hint: No free lunch rule: Training set and testing set come from the same distribution Need to make some assumptions or bias Note: - Plagiarism is defined as, the use of another person’s original (not common-knowledge) work without acknowledging its source. Thus plagiarism includes, but is not limited to: o copying in whole or in part, a picture, diagram, graph, figure, program code, algorithm, etc. and using it in your work without citing its source o using exact words or unique phrases from somewhere without acknowledgement o putting your name on a report, homework, or other assignment that was done by someone else Students are expected to familiarize themselves with how to avoid plagiarism. I encourage students to collaborate on assignments, however what this means is that you can work together to decide on solution strategies, discuss what should be included and how answer should be organized, etc., but you may not copy answers in whole or in part. Plagiarism detection software may be used to check the originality of your submissions, so do your own work.
Posted on: Wed, 28 Aug 2013 19:56:32 +0000
Trending Topics
Recently Viewed Topics
© 2015