Welcome to
Deep Learning

Welcome!

Introduction:

Welcome to CSCI 1470/2470! Over the past few years, Deep Learning has become a popular area, with deep neural network methods obtaining state-of-the-art results on applications in computer vision (Self-Driving Cars), natural language processing (Google Translate), and reinforcement learning (AlphaGo). These technologies are having transformative effects on our society, including some undesirable ones (e.g. deep fakes). This course intends to give students a practical understanding of how Deep Learning works, how to implement deep neural networks, and how to apply them ethically. We introduce students to the core concepts of deep neural networks, including the backpropagation algorithm for training neural networks, as well as specific operations such as convolution (in the context of computer vision) and word embeddings and recurrent neural networks(in the context of natural language processing). Throughout the lectures, labs, and assignments, we emphasize and require students to think critically about potential ethical pitfalls that can result from mis-application of these powerful models. The course is taught using the Tensorflow deep learning framework.

Course Objectives:

By the end of this course, you will be able to:
  • Learn about the fundamental algorithms that underly all modern deep learning models.
  • Implement different types of deep learning models in Tensorflow.
  • Think critically about using a deep learning model for a task and its potential societal impact.
  • Collaborate with classmates on a team project to apply deep learning models to task of your choice.
  • Communicate your findings (both positive and negative results are encouraged) through presentations.

Prerequisites:

  • A basic programming course: (CSCI 0150, 0170 or 0190)
  • A linear algebra course: (CSCI 0530, MATH 0520 or 0540)
  • A stats / probability course: (CSCI 0220, 1450, 0450, MATH 1610, APMA 1650 or 1655)
Exceptions may be possible for those missing one of these prerequisites if (a) the student has taken another course which covers similar material, or if (b) the student will be concurrently taking the prerequisite. If either of these situations applies to you, use the “Request Override” feature in Courses@Brown to request an override code (and explain why you believe your situation merits one).

Assessment of learning:

There are six assignments, weekly labs, and a final project. The grade breakdown for this course is as follows:
  • 65% - Assignments (programming + written components)
  • 15% - Weekly Labs
  • 20% - Final Project

Textbook:

None required. Students are encouraged to refer to the following textbook, which is available online:
  • Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron Courville.

Assignments:

Assignments consist of a programming component (e.g. implementing and training some neural network model) as well as a written component. The written component contains both algorithmic conceptual questions as well as an open-ended ethics question that requires students to reflect on societal ramifications of the machine learning models they are building in the programming portion of the assignment. Answers to the written questions are due before the programming component deadline, i.e. there are two separate deadlines: one for the written component, followed by one for the programming component.

Late Policy: Each of the assignments is due by 11:59pm AoE (Anywhere on Earth) Time on the given due date — thus, you have until the end of the day listed as the due date to complete the assignment. You will have four (4) free late days to use towards all but the last project. These late days can be used on code or conceptual assignments. After your late days are expended you will lose 10% of your project grade for each extra day your handin is delayed. At the end of the term, we will calculate how to best divvy out your late days to best help your final grade. There is no need for you to do anything special for this, other than keep track of submission dates for your own records. Note that because we do this, use of late days will not be reflected in the initial grade report for your assignments.

Exceptions to the Late Policy: Sometimes there are special circumstances during the semester that result in exceptions to this late policy. All such circumstances require an official note from the Deans. In general, they only provide support notes on behalf of students who are experiencing disruptive medical or personal circumstances, including those related to Title IX situations, that affect their ability to do academic work in a timely way. If you believe an extension is warranted for you, and you have a note from the Deans, please notify the instructor. What you should not do:

  • Do not send a dean’s note to a UTA/HTA. These often contain sensitive personal information, the receipt of which by anyone other than the instructor is a FERPA violation.
You should manage other special circumstances such as interviews, personal travel or extra-curricular factors using the late day policy above.

Submission Mistakes:

Every year, a few students suffer from submission mistakes: accidentally submitting the wrong version of their code, re-submitting a broken version which overwrites a previous correct solution, etc. Since we are using Github Classroom, all student assignment work will be stored in Github repositories. If you have a submission mistake that cannot be resolved before the deadline, inform the HTAs, and include a link to a commit in your assignment repository that represents the code you wish you had submitted. Note that this commit must have a timestamp from before the assignment submission deadline. Handling such exception cases adds overhead to an already work-intensive grading process on our part. Thus, while your first mistake carries no grade penalty, subsequent mistakes will cost you 10% of your assignment grade. In other words, it’s in your best interest to be thorough with your code submissions.

Labs:

Students will attend one two hour-long weekly lab slot assigned to them. It is expected that students should be able to complete the exercises for a given lab, and get them checked off by the lab TA, during this assigned slot. Lab materials for each week will be released online at the beginning of the week; students are welcome to look at and beginning working on the lab in advance of their lab slot, if they wish. If the student does not complete the lab and/or get it checked off during their slot, they can go to another lab section that week to get it checked off (without penalty). If a student misses their assigned lab slot, they can also go to another lab section to get the lab checked off. However, this will incur a lab grade penalty (e.g. a 3/4 instead of a 4/4 for that lab). Students will be permitted to miss their assigned lab slot three (3) times without incurring this penalty. In all cases, a lab must be checked off at latest by the student’s assigned lab slot the following week. If a student knows well in advance that they will not be able to make their assigned lab slot but will be able to make another slot that week, the student may message their lab TA at least 48 hours in advance to be temporarily switched into a different lab slot for that week. This will not incur a late lab penalty.

Lab collaboration: Students may complete labs in groups of two, if they wish. In this case, both students must be present for the TA to check them off, and the TA will ask check-off questions of both students

Final Project:

All students will be required to complete a final project in groups of 2-4. At minimum, this should entail re-implementing the methods described by a recent deep learning research paper. Expectations are higher for 2470 students and 1470 capstone students; see the relevant sections below for more information. More details will be provided during the course.

Deep Learning Day: The class’s collective final project efforts culminate in “Deep Learning Day,” an end-of-semester celebration consisting of poster sessions and oral presentations organized as a day-long symposium. Details of the day will be shared during the course.

How can you do well?

This class has a high-level course load and you can ensure your success in it by doing the following:

  • Regularly attending classes and lab sections. Participation in class discussions is highly encouraged.
  • Completing and turning in all assignments on time.
  • Equally contributing to the final project and clearly presenting your project.

Differences Between 1470 and 2470?

Both sections of this course share the same course staff, lectures, labs, and assignments. 1470 is open to everyone who meets the prerequisites. 2470 is a graduate-level course and is intended for graduate students; registration is gated accordingly. The instructor reserves the right to admit undergraduates into 2470 at her discretion.

In terms of required coursework, the two sections have the following differences:

Assignments (Programming): Every programming assignment has a quantitative metric threshold that all students’ code is expected to meet (e.g. test-set accuracy). The threshold for 2470 students is higher. Reaching this higher threshold may require more advanced model architectures, the implementation of which may require reading additional reference material.

Assignments (Written): 2470 students will be required to answer additional written questions. These may be mathematical (e.g. proving/deriving some property of a model/algorithm) or conceptual in nature (e.g. requiring reading a related paper and connecting its findings to what we have discussed in class).

Final Project: 2470 students will be required to go above-and-beyond re-implementing an existing research paper: rather, they will be required to attempt a novel research project. This could involve investigating a new model architecture to solve an existing problem, adapting methodology from a different problem to solve an existing problem, or defining and solving a new problem of interest.

Programming:

All programming in this course will be done in Python, primarily using numpy and the Python Tensorflow API. Labs will be provided as Google Collab notebooks. Since this is not a software engineering course, we won’t be enforcing stringent style guidelines, but you should write so that someone who isn’t a Python wizard will be able to understand what your program is doing (add plenty of comments, break up code into smaller functions, i.e. apply basic common sense). If you turn in a partially-functional assignment and we can’t tell what you were trying to do, we’ll probably be very grumpy about giving partial credit. As that translates to an official policy, so long as your code produces the expected output(s) and adheres to any specific project restrictions (runtime, etc.) then you will not lose points for poor design or coding practices. However, as this is not a software design course, it is not the responsibility of the TAs to attempt to understand the intentions underlying confusing code. If it is not fully clear what you were trying to do in the implementation of a partially-functional assignment (i.e. not all of the output is as expected) then partial-credit will be given sparingly and at our discretion.

Time Requirements:

In addition to 3 hours per week in class, you will probably need 1 hour of help from a UTA, 2 hours for lab, 2 hours of reading and review to solidify your grasp of the material, and 6 hours for the assignments. (182 hours/semester)

Capstone:

This course may be used as a capstone course for an Sc.B. degree. Bring a copy of the capstone form to the instructor after class or during office hours. All 1470 students who use the course as a capstone will be expected to do fulfill the same final project requirements as students taking 2470. Also please fill the capstone declaration form.

Students with Special Needs:

Brown University is committed to full inclusion of all students. Please inform me early in the term if you have a disability or other conditions that might require accommodations or modification of any of these course procedures. You may speak with me after class or during office hours. For more information, please contact Student and Employee Accessibility Services at 401-863-9588 or SEAS@brown.edu. Students in need of short-term academic advice or support can contact one of the deans in the Dean of the College office.

Diversity Statement:

This course is designed to support an inclusive learning environment where diverse perspectives are recognized, respected and seen as a source of strength. It is our intent to provide materials and activities that are respectful of various levels of diversity: mathematical background, previous computing skills, gender, sexuality, disability, age, socioeconomic status, ethnicity, race, and culture.

Academic Integrity and Collaboration Policy:

Academic dishonesty will not be tolerated. This includes cheating, lying about course matters, plagiarism, or helping others commit a violation. Plagiarism includes reproducing the words of others without both the use of quotation marks and citation. Students are reminded of the obligations and expectations associated with the Brown Academic and Student Conduct Codes.

Discussion of course material with your classmates is both permitted and encouraged. However, showing, copying, or other sharing of actual code or verbatim answers to written questions is forbidden. This includes publishing projects on Github or any other public platform. This policy will be enforced.

One of the ways that we enforce the collaboration policy is by running MOSS on all code submissions. For those who are new to the department or otherwise unfamiliar with it, MOSS (short for "measure of software similarity") is a software tool which detects similarities between pieces of code. The course staff manually examines the output of MOSS to look for cases where two students’ submitted code is similar in such a way that it is sufficiently unlikely for them to have independently produced it (i.e., an instance of cheating or code plagiarism).

Regret clause: We recognize that college is stressful, and sometimes that stress leads to panicked, bad decision making near deadlines. We are also aware that this problem is likely to be exacerbated by the unusually stressful period of history through which we are all living right now. In the spirit of this understanding, we are implementing a "regret clause" for the course this year. If a student submits an assignment that they know to contain plagiarized material (either code or answers to written questions), that student may invoke the regret clause by informing the instructor of the plagiarized material within 48 hours of the assignment submission deadline. The instructor will ask to meet with the student, and the outcome of this meeting can vary. On a first invocation of the regret clause, the instructor will exercise her discretion, applying a penalty up to receiving zero credit on the plagiarized portion of the assignment. On subsequent invocations, all plagiarized portions of the assignment will automatically receive zero credit. In all cases, the occurrence of plagiarism will be kept confidential by the instructor.

If the course staff detects a plagiarized submission, and the student has not invoked the regret clause, then we will write up a formal plagiarism case to be heard by the Academic Code Committee. These hearings can result in a variety of outcomes, including immediately failing the course and receiving a “Directed NC” (i.e. a form of NC that does show up on your transcript, along with a note that the NC was given for an academic code violation).

Accommodations:

Brown University is committed to full inclusion of all students. Please inform the instructor if you have a disability or other condition that might require accommodations or modification of any of these course procedures. You may email the instructor, come to office hours, or speak with her after class, and your confidentiality is respected. We will do whatever we can to support accommodations recommended by SEAS. For more information contact Student and Employee Accessibility Services (SEAS) at 401-863-9588 or SEAS@brown.edu. Students in need of short-term academic advice or support can contact one of the deans in the Dean of the College office.

Mental Health:

Being a student can be very stressful. If you feel you are under too much pressure or there are psychological issues that are keeping you from performing well at Brown, we encourage you to contact Brown’s Counseling and Psychological Services (CAPS). They provide confidential counseling and can provide notes supporting extensions on assignments for health reasons.

Incomplete Policy:

We expect everyone to complete the course on time. However, we certainly understand that there may be factors beyond your control, such as health problems and family crises, that prevent you from finishing the course on time. If you feel you cannot complete the course on time, please discuss with the instructor the possibility of being given a grade of Incomplete for the course and setting a schedule for completing the course in the upcoming year.

Acknowledgments:

Thanks to Tom Doeppner and Laura Dobler for the text on accommodation, mental health, and incomplete policy. This document was adapted from earlier versions written by Daniel Ritchie and Chen Sun.

Professor: Ritambhara Singh ()
Location: Solomon DECI (Room 101)
Time: MWF 12:00pm – 12:50pm

Links to Edstem and Gradescope can be found in Canvas

Canvas: Link

Instructor Office Hour: Tuesday 2pm - 4pm

Course Missive / Syllabus
Feedback Form
Capstone Declaration Form

Contact:
Course staff:
HTAs + Professor: