Sittyba
Fundamentals of Machine Learning (CS-UA 473)
Spring Semester 2025
§0.0 Purpose, design & philosophy (PDP): As data and computational resources become ever more abundant, the ability to leverage both has an increasingly transformational impact on economy, society and civilization, from prediction to generative AI. “Machine Learning” is an umbrella term for the algorithms, tools and approaches that drive this development. This class is a survey course intended to give an overview of all major flavors of Machine Learning that are in common use in the first quarter of the 21st century. Importantly, we will place a particular emphasis on understanding the foundations that machine learning algorithms rest on, as we enter the 4th age of human development. The ultimate purpose of this class is for you to be able to apply these fundamental machine learning approaches to solve real world problems both with confidence and competence.
§1.0 Instructor: Pascal Wallisch, PhD [teaches the lecture] Office: 60 Fifth Avenue, Room 210
Phone: (212) 998-8430
Email: intro2MLnyu@gmail.com
Office hours: Tu 2.15-3.15 pm (Walk-ins welcome, first come, first serve - take a fox stick)
We 1.00-2.00 pm (Walk-ins welcome, first come, first serve - take a fox stick) Th 3.00-4.00 pm (Walk-ins welcome, first come, first serve - take a fox stick)
§1.1 Teaching Assistants (email:[email protected]):
Course assistant [teaches the lab]: Umang Sharma. OH: Thu 12.30-1.30pm in 60 5th Ave, Room 402
|
Tutor [teaches one on one]: Hamza Alshamy. Schedule one-on-one sessions viaCalendly
|
Section leader [teaches the recitations]: Zhe Zeng. OH: Fr 12.00-1.00 pm, Room 340 in 60 5th Ave
|
Graders [grade assignments]: Several, anonymous, no contact (teaching vs. grading firewall)
|
§1.2 Lecture times: Mo & We 11:00 am - 12:15 pm
§1.3 Lecture space: GCASL, C95 mirrored inhttps://nyu.zoom.us/j/93881706252
§1.4 Session content: There are 3 kinds of sessions per week. On Monday, lectures introduce new course content each week, focusing on high level goals, concepts and algorithms. (Usually) on Wednesday, the lab focuses on the practical implementation of the lecture content in code, using real and synthetic data. On Friday, the recitation section focuses on implementation and practice of course materials. Sometimes, we will also feature guest speakers who will provide an industry perspective on class concepts.
§1.5 Section: Fridays, 9.30 – 10.45 am and 2.00-3.15 pm in 31 Washington Pl (Silver), Room 405
§1.6 Prerequisites: Linear Algebra, Data Structures, Probability & Statistics
§1.7 Scope: 0 to 1. Language of instruction is Python, we index from 0.
§1.8 Materials (none of these is required, they are recommended depending on your background):
Concepts: “Pattern Recognition and Machine Learning”, by Bishop
Linear Algebra: “Linear Algebra and Learning from Data” by Strang
Math: “Mathematics for Machine Learning” by Deisenroth, Faisal & Ong
Coding: “Introduction to Machine Learning with Python” by Müller & Guido
Machine Learning overview: “Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems ” by Geron
§2.0 Course grading: The total grade is calculated as follows (out of 256 points) :
A) After Action Appraisals (12) 1 point / AAA 12 points total
B) Basic course logistics quiz (1) 4 points / quiz 04 points total
C) Capstone project (1) 64 points/project 64 points total
D) Deceptive AI output (1) 4 points / output 04 points total
E) Exit survey (1) 4 points / survey 04 points total
F) Final interview & emergency skills test (1) 64 points / FIST 64 points total
G) Groundstone survey 4 points / survey 04 points total
H) Homeworks (5) 20 points / HW 100 points total
Total 256 points
§2.1 Grade cutoffs:
A
|
243-256
|
B+
|
220-229
|
C+
|
190-199
|
D+
|
150-169
|
F
|
64-127
|
A-
|
230-242
|
B
B-
|
210-219
200-209
|
C
C-
|
180-189
170-179
|
D
|
128-149
|
I
|
0-63
|
§2.2 Attendance and Participation: You are responsible for the material covered in this course. Consistent attendance is critical, and whereas all lectures will be recorded, this class has been optimized for live attendance/performance. You’ll get the most out of it that way. To incentivize attendance, we assign an attendance and participation grade with the AAA assignments. There are 14 weeks, and you need to complete 12 of these AAA assignments to get a full participation score. Each Wednesday, we will open up an assignment (“After Action Appraisal“ – AAA) on Brightspace. To avoid confusion, this assignment needs to be completed BEFORE the lecture on Monday of next week. By completing this reflection and digestion assignment, you affirm that you engaged with the class sessions. Slides and code are provided to aid note-taking. They are no substitute for attending the actual class. Basing AAA on slides instead of class attendance is an academic integrity violation.
§2.3 Homeworks: Are designed to build skills and conceptual proficiency. There are no shortcuts. Immersion is key. Thus, there are 6 assignments which are due every few weeks. Please allow yourself enough time to complete them by getting started early. Note that whereas there are 6 homeworks, we will only count the scores on the highest 5 towards your course grade. In addition, each homework contains some extra credit questions, which counts towards the homework grade.
§2.4 Capstone project: This will be something that – hopefully – sparks joy and that ties together the skills you learned in this class. We’ll release a spec sheet what it entails at a suitable time in the course, around April 1st. This project will allow you to gauge whether you enjoy solving problems with Machine Learning methods and whether the class imparted the skills to do so competently.
§2.5 FIST: Whereas we anticipate – and even encourage – you to use generative AI (like chatGPT,
Github Copilot or Sparrow) to do the weekly assignments and the capstone project, you need more
than just be good at prompting the AI to succeed in this field. For instance, during a technical
interview. There are also skills someone claiming ML expertise is just expected to have on tap,
particularly in an emergency. For scalable realism, we simulate these demands in a final, cumulative and comprehensive test, asking true/false questions with a modest attempt & incorrect (a&i) penalty. For peace of mind, you can bring any notes you want, but *all* electronic devices (computer, iPad,
phone, smartwatch, etc.) are banned. Note: As this is an in-person final that happens during finals week, be sure to make travel arrangements accordingly. If you miss it, you will get an incomplete.
§2.6 Deceptive AI output: Prompt a generative AI to say something about ML that is incorrect.
§2.7 Surveys and quizzes (B, E, G): These are low stakes assignments that will help us calibrate the class, fine-tuning them to match for your needs, wants, and competencies optimally.