CS 271 Syllabus

San Jose State University
Department of Computer Science
CS 271, Topics in Machine Learning, Fall 2020

Course and Contact information
- Instructor: Mark Stamp
- Office Location: MH 216
- Email: mark.stamp@sjsu.edu
- Text: for oh ate fore owe to 10-five-too
- Office hours: Tuesday & Thursday, 10:15am - 11:00am, or by appointment
- Class Days/Times: Tuesday & Thursday, 9:00am - 10:15am
- Classroom: Online
- Prerequisites: CS 149

Course Description
- Topics in machine learning. The following machine learning techniques and related topics are covered: hidden Markov models (HMM), principal component analysis (PCA), support vector machines (SVM), clustering, data analysis, backpropagation, convolutional neural networks (CNN), recurrent neural networks (RNN), long short-term memory (LSTM), residual networks (ResNet), generative adversarial network (GAN), Word2Vec, transfer learning, ensemble techniques, and related topics. Illustrative applications of each of the major topics are provided, with most of the applications drawn from the field of information security. Additional machine learning topics will be covered as time permits.
Learning Outcomes
- The focus of this course will be machine learning, with illustrative applications drawn primarily from the field of information security. After completing this course students should have a working knowledge of a wide variety of machine learning topics, and have a good understanding of how to apply such techniques to real-world problems.
Textbook and Readings
- The primary text will be Machine Learning with Applications in Information Security (https://www.crcpress.com/Introduction-to-Machine-Learning-with-Applications-in-Information-Security/Stamp/p/book/9781138626782), by Mark Stamp, published by Chapman Hall/CRC in 2017. This book covers several machine learning techniques in detail, and includes a number of illustrative applications. Many of the applications are from information security, including a variety of topics related to malware, intrusion detection (IDS), image spam, and classic cryptanalysis, among others.
- Additional required reading material:
  - A revealing introduction to hidden Markov models, at https://www.cs.sjsu.edu/~stamp/RUA/HMM.pdf
  - Boost your knowledge of AdaBoost, at https://www.cs.sjsu.edu/~stamp/RUA/ada.pdf
  - Deep thoughts on deep learning, at https://www.cs.sjsu.edu/~stamp/RUA/ann.pdf
  - Alphabet soup of deep learning topics, at https://www.cs.sjsu.edu/~stamp/RUA/alpha.pdf
- Additional relevant material:
  - PowerPoint slides at http://www.cs.sjsu.edu/~stamp/ML/powerpoint
  - Current semester lecture videos are available at http://www.cs.sjsu.edu/~stamp/ML/lectures/CS271_Fall20/. If you are asked to login to access the videos, both the username and password are "infosec". Note: The instructor hereby gives students permission to record his lectures (audio and/or video). At least with respect to this class, your instructor has nothing to hide.
  - Class-related discussion will be posted on Piazza at https://. You are strongly encouraged to participate by asking questions, as well as by responding to questions that other students ask. At the start of the semester, you should receive an email asking you to join this discussion group—if not, contact your instructor via email.
- The applications parts of this course are essentially self-contained, but for additional background information on the security-related topics, the following resources are recommended.
  - Computer Viruses and Malware, John Aycock, Springer 2006. Many of the applications we discuss are related to malware. Aycock's book is easy to read and in spite of being fairly old, it provides a good foundation for malware research.
  - Information Security: Principles and Practice, Mark Stamp, Wiley 2011. If you have not taken CS 265, you should do so. In any case, you can refer to this fine book if you have questions about security-related topics during this course.
  - Open Malware (at http://www.offensivecomputing.net/) includes a large collection of samples of live malware.
  - VX Heavens (at http://vx.netlux.org/) is a source for "hacker" type of information on viruses. Malware samples are also available.
  - Journal of Computer Virology and Hacking Techniques (at http://www.springer.com/computer/journal/11416) is a journal for malware-specific research papers. There are also several good conferences that focus on malware and/or machine learning applications in information security.
  - Recent masters project reports (at http://www.cs.sjsu.edu/~stamp/cv/mss.html#masters). Most of these projects involve applications of machine learning to malware or other topics in information security.
Course Requirements and Assignments
- SJSU classes are designed such that in order to be successful, it is expected that students will spend a minimum of forty-five hours for each unit of credit (normally three hours per unit per week), including preparing for class, participating in course activities, completing assignments, and so on. More details about student workload can be found in University Policy S12-3 at http://www.sjsu.edu/senate/docs/S12-3.pdf.
- Approximate Schedule
  - Week 1 --- Introduction and overview
  - Week 2 --- Hidden Markov Models
  - Week 3 --- Data Analysis
  - Week 4 --- Profile Hidden Markov Models
  - Week 5 --- Principal Component Analysis
  - Week 6 --- Support Vector Machines
  - Week 7 --- Clustering
  - Week 8 --- Selected topics from k-Nearest Neighbor, Neural Networks, Boosting/AdaBoost, Random Forests, Linear Discriminant Analysis
  - Week 9 --- Backpropagation
  - Week 10 --- Convolutional neural networks
  - Week 11 --- Recurrent neural networks & LSTM
  - Week 12 --- Generative adversarial network (GAN) & ResNet
  - Week 13 --- Word2Vec
  - Week 14 --- Transfer learning, ensemble techniques, TF-IDF, dropout (and other regularization techniques), etc.
  - Week 15 --- Project presentations
- Homework is due typewritten (include source code, but not executable files) by class starting time on the due date. Each assigned problem requires a solution and an explanation and work detailing how you arrived at your solution. Cite any outside sources used to solve a problem. When grading an assignment, your instructor may ask for additional information. Note that a subset of the assigned problems will typically be graded.
  
  Homework must be submitted via email before the start of class on the due date. Be sure to have an extra copy of your homework with you in class, and be prepared to discuss your solutions. Your written solutions must be in a pdf file. Submit any source code or other attachments in separate files (i.e., no code in the solution itself). You must provide enough discussion of your solution so that the grader can understand your solution, and so that the grader can be sure that you understand your solution. Put your written solution and any relevant source code in a folder named "yourlastname". Then zip your homework folder and submit the file yourlastname.zip via email to CS271.fall20@gmail.com. The subject line of your email must be of the form:
```
     CS271HMK assignmentnumber yourlastname last4digitofyourstudentnumber 
```
  The subject line must consist of the four identifiers listed. There is no space within an identifier and each identifier is separated by a space.
  - Assignment 0: Due Tuesday, August 25
    As with all assignments, this is due prior to the start of class.
    1. Read A Revealing Introduction to Hidden Markov Models (at https://www.cs.sjsu.edu/~stamp/RUA/HMM.pdf) and do the following.
      1. Briefly (1 paragraph) summarize how an HMM is trained.
      2. How is a trained HMM used to score a sequence?
      3. Very briefly explain how an HMM and dynamic program differ.
      4. Why is it necessary to scale the values of the matrices when training an HMM?
    2. Read the article "Models will run the world" (your instructor will post a softcopy of the article by the first day of class) and do the following.
      1. In one paragraph, summarize the authors' main points.
      2. Write second paragraph discussing what you most agree with and anything that you disagree with in this article.
  - Assignment 1: Due Thursday, September 3
    Chapter 2, problems 1, 2, 3, 10. For problem 10, you must use HMM code that you have written yourself, based on the pseudo-code given on pages 24 through 28 of the textbook. Make your code as efficient as possible, since you will need to train a large number of models for the next assignment.
  - Assignment 2: Due Tuesday, September 15
    Chapter 2, problems 14 and 15 a) through 15 d). You can use the HMM code found at http://www.cs.sjsu.edu/~stamp/RUA/HMM_ref_fast.zip. If you use your own code to successfully solve the problems, you will earn extra credit.
  - Assignment 3: Due Tuesday, September 22
    Chapter 8, problems 1, 4, 5, 6, 7, 8, 10.
  - Assignment 4: Due Thursday, October 1
    Chapter 4, problems 1, 4, 11, 13, 17. For problem 13, you will use the same same eigenvectors in part b) as you use in part a).
  - Assignment 5: Due Thursday, October 15
    Chapter 5, problems 3, 4, 9, 10, 11, 12, 15.
  - Assignment 6: Due Thursday, November 5
    Problems 1, 2, 4, 6, 8, from the neural networking tutorial at https://www.cs.sjsu.edu/~stamp/RUA/ann.pdf.
  - Assignment 7: Due Thursday, November 19
    Chapter 6, problems 2, 4, 5, 7, 8, 13, 16. For problem 13, use your EM program from problem 8. Also, the data points for problem 16 can be found at http://www.cs.sjsu.edu/~stamp/ML/files/dbscan.txt.
  - Assignment 8: Due Thursday, December 3
    Chapter 7, problems 1, 4, 7, 11.
    Problem 2 from the AdaBoost tutorial at https://www.cs.sjsu.edu/~stamp/RUA/ada.pdf
    Note that you must write and use your own AdaBoost program to solve this problem.
  - Assignment 9: Due Varies
    You are required to attend at least one of the Master's defense talks listed at http://www.cs.sjsu.edu/faculty/stamp/defenses/fall20.html You can attend a second defense for extra credit (actually, you can attend as many as you want, but more than two will not yield extra extra credit).
  - Assignment 10: Due TBD
- NOTE that University policy F69-24 at http://www.sjsu.edu/senate/docs/F69-24.pdf states that "Students should attend all meetings of their classes, not only because they are responsible for material discussed therein, but because active participation is frequently essential to ensure maximum benefit for all members of the class. Attendance per se shall not be used as a criterion for grading."
Grading Policy
- Test 1, 100 points. Date: Tuesday, October 20.
- Homework, quizzes, class participation and other work as assigned, 100 points. A subset of the assigned problems will be graded.
- Machine Learning Project, 100 points. You must obtain approval for your project proposal from me (via email) by the close of business on October 6, and you must be prepared to give a brief presentation of your proposed topic on October 8. A written project report is due December 1 and project presentations will begin then.
- Final, 100 points. Date: Wednesday, December 9 from 7:15am to 9:30am. The official finals schedule is here: https://www.sjsu.edu/classes/final-exam-schedule/fall-2020.php
- Semester grade will be computed as a weighted average of the major scores listed above.
- No make-up tests or quizzes will be given and no late homework or project (or other work) will be accepted.
- Grading Scale:
  
  Percentage Grade
  92 and above A
  90 - 91 A-
  88 - 89 B+
  82 - 87 B
  80 - 81 B-
  78 - 79 C+
  72 - 77 C
  70 - 71 C-
  68 - 69 D+
  62 - 67 D
  60 - 61 D-
  59 and below F
- Note that "All students have the right, within a reasonable time, to know their academic scores, to review their grade-dependent work, and to be provided with explanations for the determination of their course grades." See University Policy F13-1 at http://www.sjsu.edu/senate/docs/F13-1.pdf for more details.
Guest Lectures
- TBD
  - Date: TBD
  - Time: TBD
  - Location: TBD
  - Title: TBD
  - Abstract: TBD
  - Short Bio: TBD
- TBD
  - Date: TBD
  - Time: TBD
  - Location: TBD
  - Title: TBD
  - Abstract: TBD
  - Short Bio: TBD
Classroom Protocol
- Keys to success: Do the homework, complete a good project, and attend class
- Wireless laptop is required. Your laptop must remain closed (preferably in your backpack and, in any case, not on your desk) until your instuctor informs you that it is needed for a particular activity
- Cheating will not be tolerated, but working together is encouraged
- Student must be respectful of the instructor and other students. For example,
  - No disruptive or annoying talking
  - Turn off cell phones
  - Class begins on time
  - Class is not over until your instructor says it is over
- Valid picture ID required at all times
- The last day to drop without a "W" grade is August 31, 2020, and the last day to add is September 8, 2020
University Policies
- Office of Graduate and Undergraduate Programs maintains university-wide policy information relevant to all courses, such as academic integrity, accommodations, etc. You may find all syllabus related University Policies and resources information listed on GUP’s Syllabus Information web page at http://www.sjsu.edu/gup/syllabusinfo/

Percentage	Grade
92 and above	A
90 - 91	A-
88 - 89	B+
82 - 87	B
80 - 81	B-
78 - 79	C+
72 - 77	C
70 - 71	C-
68 - 69	D+
62 - 67	D
60 - 61	D-
59 and below	F