đź“– Syllabus

Welcome to EECS 245! We’re so excited to have you in the course. Make sure to read the syllabus carefully so that you understand what is expected of you.

Overview

Instructor

See the 👩‍🏫 Staff page for contact information.

Content

Linear algebra, calculus, and probability form the basis of modern machine learning and artificial intelligence. This course will introduce linear algebra from scratch by focusing on methods and examples from machine learning. It will give students strong intuition for how linear algebra, calculus and probability are used in machine learning. While the course is primarily theoretical, we’ll look at practical applications involving real data in Python each week, so that students are able to apply what they’ve learned.

This course aims to provide:

  • Enough rigor to ensure students can succeed in more theoretical ULCS courses (like EECS 445).
  • Enough context, motivation, and applications for students to see why linear algebra is useful in machine learning, throughout the entire course.

Credit

This course is worth 4 credits, and is currently approved to count towards the linear algebra requirement for:

  • Computer Science majors (CS-Eng and CS-LSA*).
  • Data Science majors (DS-Eng and DS-LSA).
  • Statistics majors.
  • Artificial Intelligence minors.

And the linear algebra prerequisite for:

  • EECS 442, EECS 445, EECS 448, EECS 474, EECS 476, and CSE 576.
  • Any 400-level STATS or DATASCI course with a linear algebra prerequisite.

*CS-LSA students don’t have a linear algebra requirement, but this course fits in the linear algebra “bucket” for the second math course; see here for details.

Who should take this course?
You should take this course if you:
  • Have not already taken a linear algebra course (if you've already taken Math 214, 217, 417, or 419, this course is not for you).
  • Plan on taking Upper-Level CS courses about machine learning and artificial intelligence (e.g. EECS 445) and/or want to gain exposure to machine learning and Python early in your academic career.

Prerequisites

  • EECS 203 or Math 116 (or Math 215, 216, 275, 285, or 295), i.e. some math course after Calculus 1.
  • Some 100-level programming class (e.g. EECS 183, ENGR 101, ROB 102, or similar).

If you’re not sure if you meet the prerequisites – or if you don’t, but are still interested in taking the course nonetheless – email the instructor.

Getting Started

The course website, eecs245.org, will contain links to all course content. There are also a few things you’ll need to do to get set up.

Websites

You’ll need to make accounts on the following sites:

  • Ed: We’ll be using Ed as our course message and discussion board. More details are in the Communication section below. If you didn’t already get an invitation to our Ed course, sign up here.

  • Pensive: Pensive is where you’ll submit your homeworks and where all of your grades will live. As is discussed below, all homeworks must be handwritten, scanned into a PDF, and submitted to Pensive. Some homeworks will contain autograded Jupyter Notebooks; these will be submitted to Pensive as well (not the EECS department autograder). If you didn’t already receive an invitation to Pensive, please let an instructor know. Pensive is like Gradescope, but it allows for AI grading of handwritten work, with human review and the ability to ask for a regrade. Not only does this enable faster and more accurate grading, but it also frees up instructor time to spend working with students one-on-one, which AI cannot do. Pensive has been approved by the College of Engineering, who ensured that it meets the necessary privacy and accessibility standards; rest assured, Pensive does not train AI models on your work.

Note that we will not be using Canvas for anything this semester (so please don’t try and send us messages on Canvas!).

Programming Environment

Some labs and homeworks will involve writing Python code in Jupyter Notebooks. As soon as you can, follow the steps in the Environment Setup page to set up your programming environment locally on your computer. Alternatively, we’ve created a Jupyter Notebook server that will allow you to access and write code directly from your browser, but it can be unreliable, so you should only use it as a last resort.

Forms

Please fill out the required Welcome Survey to tell us a bit more about your background and whether you need alternate exams no later than Friday, September 11th, along with Homework 1.

Communication

This semester, we’ll be using Ed as our course message board. You will be added to Ed automatically; use the invite link in the section above if you weren’t added.

If you have a question about anything to do with the course — if you’re stuck on a problem, didn’t understand something from lecture, want clarification on course logistics, or just have a general question about machine learning — please ask on Ed! We know that generative AI tools can answer your questions, and while we encourage you to use those tools responsibly as well, part of the benefit of attending a school like Michigan is the ability to form connections with your instructors and peers. We only ask that if your question includes some or all of an answer (even if you’re not sure it’s right), please make your post private so that others cannot see it. You can also post anonymously to other students if you prefer.

Course staff will regularly check Ed and try to answer any questions that you have. You’re also encouraged to answer questions asked by other students. Explaining something is a great way to solidify your understanding of it!

Lectures

Lectures will be held in-person on Tuesdays and Thursdays from 10:30AM-12PM in 1010 DOW. Lectures will be recorded and linked directly on the course homepage. Recordings will be made publicly available so that students who are not enrolled – including students not at Michigan – can benefit from the recordings. As part of your participation in this course, you may be recorded (e.g. if you answer a question). If you do not wish to be recorded, please email the instructor to discuss alternate arrangements.

Course Notes

Course notes have been written specifically for EECS 245, and will be your main resource in this class. These can be found at notes.eecs245.org. We are constantly refining the notes, and will link them directly underneath the relevant lecture once they are ready.

These notes aren’t optional readings, they’re mandatory. It is nearly impossible to succeed in this course without reading them regularly. See the Advice section of the notes for testimonies from previous students.

If you find any typos or ways to improve the notes, please fill out this form.

Live lectures will closely follow the notes. Instead of presenting using slides in lecture, we will start with a blank document on a tablet, and write out the lecture content in real time, allowing us to move at a pace that is appropriate for the class and to focus on the high-level ideas (details can be found in the notes). Anything we write in live lectures will be posted as a PDF on the course website after lecture.

Attendance

Lecture attendance is taken, and is optionally worth 5% of your overall course grade. How is this assessed?

  • There will be 26 regular lectures throughout the term; each lecture you attend will earn you 1 “lecture point”, up to a maximum of 22 lecture points. (Lectures on the days of the two midterms will not count for lecture points.)
  • Your lecture score will be the number of lecture points you earn out of 22. This means you can miss up to 4 lectures for any reason (late add, extenuating circumstances, etc.) and still earn a full lecture attendance score.

5% of your overall grade in the class either comes from your attendance score OR your Final Exam score – whichever is higher. This policy is designed to encourage you to come to lecture as often as you can. We will do our best to make lectures engaging, active, and well worth your time.

We will use PollEverywhere to take attendance. It is an academic integrity violation to check into attendance if you are not physically present in lecture, so please do not consider doing that! If you have a regularly scheduled conflicting class, let Suraj know and we can discuss alternatives.

Labs

Labs will be held in-person on Wednesdays at several times. You can attend any lab section on any given week, regardless of which section you’re enrolled in. Exact times and rooms will be posted on the Calendar page.

Labs are meant to provide hands-on practice with the recent lecture material and preparation for the upcoming homeworks and exams.

Each lab session will have an accompanying physical worksheet that is distributed in-person at the start of lab and meant to be completed during the live lab session. Each worksheet is broken into several activities; some activities will involve writing math on the paper worksheet, and others will involve writing code in a Jupyter Notebook and either showing your TA a working solution or submitting it to the autograder.

To earn credit for a lab, you must make an effort to attempt all “required” activities and show the lab TA an almost completed worksheet by the end of the lab session. This means that there is no option to receive credit for labs remotely. Worksheets will usually include additional practice problems for you to work through after lab.

There will be 12 labs in total, and labs are worth 10% of your grade. Each week you attend and submit the lab will earn you 1 “lab point”, up to a maximum of 9 lab points. Your lab score will be the number of lab points you earn out of 9. This means you can miss up to 3 labs for any reason (late add, extenuating circumstances, etc.) and still earn a full lab score. This flexibility is provided in lieu of allowing for remote submissions, since we really believe in the power of working on problems in-person and on-paper with other students. Details can be found in the Grades section below.

Lab worksheets and solutions will be posted on the course website after all lab sessions have concluded.

Homeworks

This class will have 12 weekly homework assignments, which are to be completed individually; homeworks are worth 15% of your grade. Homeworks will usually be released on Friday afternoons and be due the following Friday at 11:59PM, though this may change in any given week. See the course homepage for the most up-to-date deadline schedule.

Homeworks must be handwritten on paper or tablet by each student individually, scanned into a single PDF, and submitted to Pensive. That is, no typing and no LaTeX, unless you have accommodations specifying otherwise. We believe that handwriting the material helps you learn it better, better prepares you for exams, and frankly is harder to cheat with AI.

Collaboration and Generative AI Policy

The majority of homework problems will involve writing solutions to math problems, but some will involve writing code in a Jupyter Notebook. The point of a homework problem is not really to “find the answer” – we know the answers, and ChatGPT can easily provide answers to any homework problem we could ask you. Rather, the point of a homework problem is to help you deeply understand the material, apply it to new problems, and build problem-solving skills along the way that will pay dividends in future courses and in your life as a graduate of the University of Michigan.

Think of it like this: driving 10 miles is a lot easier than running 10 miles, and in many cases, driving makes total sense. But, if you’re trying to exercise, driving 10 miles won’t really help you achieve your goals. If you buy that analogy, just think of driving as “copying the answers from someone else” and running as “thinking through problems yourself” – both may earn you the same score on homeworks, but only the latter will help you in the long run.

With all of that in mind, you can – and are actually encouraged to! – talk to other students in the class about the problems and discuss solution strategies. Talking and debating about challenging ideas is a great way to solidify your own understanding of the material. That said, you should not share any written communication. You can tell someone how to approach a homework problem if they’re stuck, but you cannot show them how to do it. One way to tell if you are respecting this boundary is to ask yourself whether your collaboration could take place over the phone. Additionally, the content of your verbal communication should involve the problem-solving strategy and approach, and you should not directly compare answers with classmates.

Similarly, you can ask generative AI tools like ChatGPT for help with concepts, because these tools can be genuinely useful. If you find a particularly useful explanation online, either on a website or from a generative AI tool, please let us know! You cannot ask generative AI tools to solve homework problems for you or to verify your answers. Doing so is only cheating yourself. Remember you will not be able to use these tools during the exams, which are worth 70% of your overall grade.

If we suspect that your homework answers are not the result of your own work, we reserve the right to interview you in-person and ask you to answer and explain similar questions, and submit a case to the Honor Council.

You may post homework-related questions on Ed, though your questions and answers should be about approaches, not answers to specific problems. If your question includes some or all of an answer (even if you’re not sure it’s right), you must make your post private so that others cannot see it. We are not able to tell you whether your answer is correct.

One final thought: Demis Hassabis, the co-founder of Google’s DeepMind AI research organization and a 2024 Nobel Prize winner, said

Those people who understand the deep technical, they’ll be able to use these tools 10 times more effectively than people who don’t have that technical knowledge. (source)

While it’s true that generative AI tools can answer any question we’d ever ask you in the course, remember that the point is to gain independent critical thinking skills. Keep this at the back of your mind throughout the semester!

Late Policy, Slip Days, and Drops

All homeworks must be submitted by 11:59PM Ann Arbor time on the due date to be considered on time. If you make a submission after the deadline, your assignment will be counted as late.

You have 8 “slip days” to use throughout the semester. A slip day extends the deadline of a homework by 24 hours. You may use up to 2 slip days on any one homework assignment. (Since lab worksheets must be submitted in-person, they cannot be submitted late, and can’t use up slip days.)

Slip days are designed to be a transparent and predictable source of leniency in deadlines. You can use a slip day if you are too busy to complete a homework on its original due date (or if you forgot about it). But slip days are also meant for things like the internet going down at 11:58PM just as you go to submit your homework. Slip days are meant to be used in exceptional circumstances, so you probably should not need to use all 8, but if you have something going on in your life that is impeding your ability to do your classwork on time, please reach out to us as soon as possible so we can work something out. The earlier you let us know that something’s going on, the more we can do to help, so please reach out.

Slip days are applied automatically at the end of the semester, and you don’t need to ask in order to use one. It’s your responsibility to keep track of how many you have left. If you’ve run out of slip days and submit an assignment late, that homework may still be graded, but you will receive a 0 on it when we calculate grades at the end of the semester.

In addition, your lowest 2 homework scores will be dropped from your final grade. This means that you can miss up to 2 homeworks for any reason and still earn a full homework score. You do not use up any slip days on homeworks you don’t submit, e.g. if you choose not to submit Homework 2, it costs you 0 slip days.

Regrade Requests

If you believe that we have made a mistake in applying the rubric shown to you on Pensive, you may submit a regrade request directly on Pensive within one week of the grades being released. If you do not submit a regrade request within one week, your original grade will be final. Part of your grade is clarity, so if your answer was mostly right but unclear you may still not be eligible for full credit.

Some homework problems will be graded automatically by the autograder. If you believe that the autograder has made a mistake in grading your homework, send the instructor an email, again, within one week of the grades being released. Note that it’s rare that something is wrong with the autograder, and if that’s the case, we’ll typically fix the necessary test cases and re-run the autograder for the entire class.

A Typical Week

To summarize the details above, here’s what a typical week looks like in EECS 245:

MondayTuesdayWednesdayThursdayFriday
 LectureLabLectureHomework \(N-1\) Due
Homework \(N\) Released

Office Hours

To get help on assignments and concepts, course staff will be hosting several office hours per week. The vast majority of office hours will be held in-person on North Campus, though we may hold a few remote office hours, depending on staff availability. See the Calendar tab of the course website for the most up-to-date schedule and directions. If you aren’t able to consistently attend at least one office hours slot a week, please email the instructor – we will try our best to move things around to get you the support you need.

In a theoretical class such as this one, it’s important to make good use of office hours. Homework assignments will be challenging, so you should plan to attend office hours at least once a week. Even if you don’t have specific questions, you will likely get a lot out of conversing about the material and hearing others’ questions. And even if you’ve mastered the material, still come and say hi – take advantage of the fact that this is a relatively small class!

Exams

This class has two Midterm Exams and one Final Exam, all of which will be administered in-person and on paper. You can view past exams at exams.eecs245.org.

ExamDate and TimeContentWeight
Midterm 1Tuesday, October 6th, 7-9PMLectures 1-10, Chapters 1-4 only20%
Midterm 2Tuesday, November 17th, 7-9PMLectures 11-19, Chapters 5-8 only20%
Final ExamWednesday, December 16th, 1:30-3:30PM~30% Lectures 1-10
~30% Lectures 11-19
~40% Lectures 20-26		30% or 35%

The specific lecture numbers above are subject to change slightly. Midterm 2 is not cumulative. The Final Exam will be cumulative, and broken into three parts:

  1. Part 1 of the Final Exam will be based on Midterm 1 content. If you score higher on Part 1 than you did on Midterm 1, we will replace your Midterm 1 score with \(\frac{\text{Part 1 score} + \text{Midterm 1 score}}{2}\).
  2. Part 2 of the Final Exam will be based on Midterm 2 content. If you score higher on Part 2 than you did on Midterm 2, we will replace your Midterm 2 score with \(\frac{\text{Part 2 score} + \text{Midterm 2 score}}{2}\).
  3. Part 3 of the Final Exam will be based on the content introduced after Midterm 2.

This “redemption policy” for Parts 1 and 2 is designed to help you boost your Midterm 1 and Midterm 2 scores if you didn’t do as well as you’d hoped. This policy can only help your grade; it can’t hurt. To be clear, all three parts of the Final Exam are required and part of your Final Exam score.

If you have conflicts with any of the exams, please let us know on the Welcome Survey. We may provide alternate exam times for students with a valid, documented conflict with a required activity in another course or official university-affiliated activity, or to help students avoid negative academic consequences when their religious obligations conflict with academic requirements.

Exams are to be completed individually, with absolutely no collaboration allowed. Any suspected violations will be reported to the Honor Council.

Grades

Weights

ComponentWeightNotes
Midterm 120%see Exams section above
Midterm 220%see Exams section above
Final Exam30% 
Homeworks15%• 12 total; 2 lowest scores dropped
• 8 slip days available to use with a max of 2 per homework
Labs10%12 total; 3 lowest scores dropped
max(Lecture Attendance, Final Exam)5%Lecture Attendance score: 26 total; 4 lowest dropped

Letter Grades

Grading for this class is not curved in the sense that the average is set at (say) a B+ and half of the class must receive a grade lower than that. If everyone does well and shows mastery of the material, everyone can receive an A (this would be awesome!). If no one does well (this is unlikely), then everyone can receive a C.

Grading for this class is curved in the sense that we do not have a pre-defined mapping from overall scores to a final GPA. There is no pre-determined score (e.g., 90% of all possible points) that earns an A or a B or a C or any other grade. To determine the final grade, we will ask questions like “Did this student master the material?”.

With that said, grades will not be any stricter than the standard grading scale, where an A+ is a 97+, A is 93+, A- is 90+, etc. For instance, the threshold for an “A” will never be higher than 93%. The median grade both semesters last year was a B+, and we expect this semester’s distribution to be similar.

Try your best not to worry about grades, and we’ll reciprocate by being fair. We’re in this together ❤️.

Student Support and Well-Being

Accommodations

If you need, or think you might need, an accommodation for a disability, please let us know during the first three weeks of the semester. Some aspects of this course may be modified to facilitate your participation and progress. As soon as you make us aware of your needs, we can work with the Services for Students with Disabilities (SSD) office to help us determine appropriate academic accommodations. SSD (ssd.umich.edu; 734-763-3000) recommends accommodations through a Verified Individualized Services and Accommodations (VISA) form. Any information you provide is private and confidential and will be treated as such.

Diversity and Inclusion

It is our intention that students from all backgrounds and perspectives will be well served by this course, and that the diversity that students bring to this class will be viewed as an asset. We welcome individuals of all ages, backgrounds, beliefs, ethnicities, genders, gender identities, gender expressions, national origins, religious affiliations, sexual orientations, socioeconomic background, family education level, ability - and other visible and nonvisible differences. All members of this class are expected to contribute to a respectful, welcoming, and inclusive environment for every other member of the class. Your suggestions are encouraged and appreciated.

Campus Resources

As a student, you may experience a range of issues that can negatively impact your learning, such as anxiety, depression, interpersonal or sexual violence, difficulty eating or sleeping, loss/grief, and/or alcohol/drug problems. These mental health concerns or stressful events may lead to diminished academic performance and affect your ability to participate in day-to-day activities.

In order to support you during such challenging times, the University of Michigan provides a number of confidential resources to all enrolled students, many of which are listed here. Some particularly useful resources include:

Acknowledgements

This course is new at the University of Michigan. With that said, many of the materials we will use are adopted from content created by countless other instructors for courses at other institutions.

  • In particular, some materials are adopted from EECS 398: Practical Data Science, which itself was based on content from DSC 10, DSC 40A, and DSC 80 at the University of California, San Diego; and Data 6 and Data 100 at the University of California, Berkeley.
  • Some homework problems have been adapted from various linear algebra courses and textbooks online and at other institutions. Some of these sources are listed at the bottom of the introduction to the course notes and on the resources page.
  • Language in this syllabus has been adopted from other courses as well, including EECS 203, EECS 280, EECS 376, and EECS 485 here at the University of Michigan, and CSE 160 at the University of Washington.

Disclaimer

While we try to do our best to plan ahead, unfortunately, sometimes circumstances do arise that necessitate a policy change. When this happens, the change will be announced, and this document will be updated with the new policy.

We appreciate any and all feedback, given that this course is new and evolving. If you’d like to provide us with anonymous feedback at any point, you can do so at this form. Thank you.

Note about Accessibility
It’s important to us that all students are able to fully engage with our course content. We’re working to ensure that all digital materials – the website, course notes, homework and lab PDFs, supplemental videos, etc. – meet WCAG 2.1 AA accessibility standards, as mandated by the federal government of all public universities in the United States. If you find any accessibility issues (hard-to-read colors, missing alt text, videos with missing captions, or anything else that makes the content difficult to access), please let Suraj know as soon as possible at rampure@umich.edu.