CSCI 4750: Machine Learning

Saint Louis University · Spring 2027 · Section 01 · CRN 23042 (undergraduate)
Mon/Wed/Fri 10:00 – 10:50 AM · Room TBA · Jan 11 – May 3 (last class meeting)
Project presentations run on three days — Fri Apr 30 and Mon May 3 (in class) and Wed May 5, 12:00–1:50 PM (final-exam slot); no written final
Instructor: Jin Zhao · Office: Ritter Hall 309 · Email: jin.zhao@slu.edu
Office hours: Mon/Wed/Fri 11:00 AM – 12:00 PM in Ritter Hall 309, and by appointment

CSCI 4750 is cross-listed with CSCI 5750-01 (graduate, CRN 23043): same meetings, same lectures and materials, same midterm. Requirements differ in the homework (see Assessment). This syllabus states the requirements for CSCI 4750 (undergraduate) students; graduate students should consult the CSCI 5750 syllabus.

Meeting type: In person. If in-person attendance ever becomes infeasible, class will meet on Zoom instead — I will announce any such change in advance on Canvas, where the Zoom link is posted.

Course Description

This course introduces the field of machine learning, with emphasis on the probabilistic models that dominate contemporary applications. You will learn how computers learn from examples and extract patterns hidden in large data sets: classification algorithms that predict discrete labels, regression algorithms that predict continuous values, and both supervised and unsupervised settings — labeled training data available or not. The course stresses the conceptual relationships between these learning problems and the statistical models and computational methods behind them.

The course is about fundamentals: every method is derived, worked by hand on small numbers, and implemented from scratch before you use a library version. It is also about now: each lecture closes by showing where its classical idea lives in 2026–27 industry and research — the softmax regression inside a language model, the nearest-neighbor search behind recommendations, the gradient boosting inside fraud detection, the diffusion models behind image generators, the reinforcement learning behind reasoning models. The arc runs from k-nearest neighbors in week one to foundation models in April, measured all semester on one shared benchmark.

The course is lecture-oriented, with four substantial programming assignments, an in-class midterm, and a final project (teams of 1–4) presented on the three presentation days.

Audience & Prerequisites

CSCI 4750 is the undergraduate listing of a course that also serves M.S. students (as CSCI 5750); it is intended for advanced undergraduates.

• Required (per the catalog): STAT 3850, CSCI 2100 (Data Structures), and MATH 2530 (Calculus III). In practice: solid Python programming; probability and statistics at the level of a full statistics sequence; multivariable calculus.
• Not assumed: linear algebra beyond the basics, NumPy, scikit-learn, or PyTorch. Lectures 3–4 are a self-contained math toolkit (linear algebra, gradients, optimization) taught for exactly what the course uses, and HW1 ships a NumPy warm-up and a setup check.

Learning Outcomes

By the end of the course, students will be able to:

1. Formulate a prediction problem as learning from data — features, labels, hypothesis class, loss, and evaluation — and explain the difference between fitting the training data and generalizing.
2. Derive and implement core supervised methods from scratch — nearest neighbors, linear and logistic regression, regularized models, Naive Bayes, decision trees — and explain how kernels, SVMs, and ensembles extend them.
3. Connect losses to likelihoods: explain maximum likelihood, MAP estimation, and Bayesian inference, and use the probabilistic view to choose and justify a model.
4. Apply unsupervised methods — k-means, Gaussian mixtures with EM, PCA, embeddings, matrix factorization — and interpret their output honestly.
5. Build, train, and debug neural networks, from backpropagation by hand to convolutional networks in PyTorch, and explain how self-supervised learning, generative models, and reinforcement learning power today’s foundation models.
6. Evaluate models rigorously — cross-validation, appropriate metrics, leakage checks, fairness criteria, and distribution shift — and recognize when a reported number should not be trusted.
7. Conduct and communicate an original ML project: formulate a question, run experiments against baselines, and present findings in a written report and a talk.

Materials

All required materials are free.

• [ISLP] James, Witten, Hastie, Tibshirani & Taylor, An Introduction to Statistical Learning with Applications in Python (Springer, 2023) — https://www.statlearning.com/ (primary text for Units 1–3; written for readers with a statistics background).
• [PML] Kevin Murphy, Probabilistic Machine Learning: An Introduction (MIT Press,
  2022. — https://probml.github.io/pml-book/book1.html (the probabilistic view, in more mathematical depth).
• [MML] Deisenroth, Faisal & Ong, Mathematics for Machine Learning (Cambridge, 2020) — https://mml-book.github.io/ (for the math-toolkit lectures and PCA/GMM).
• [UDL] Simon Prince, Understanding Deep Learning (MIT Press, 2023) — https://udlbook.github.io/udlbook/ (primary text for Unit 4).
• [S&B] Sutton & Barto, Reinforcement Learning: An Introduction, 2nd ed. (2018) — http://incompleteideas.net/book/the-book-2nd.html (the two RL lectures).
• Papers and posts for the “Now” segments and the modern-ML lectures (listed in the class schedule).
• Reference courses if you want another angle: Stanford CS229, CMU 10-601, Berkeley CS189.

Compute: assignments are sized for Google Colab; nothing requires a paid account. A GPU helps only in HW4 and in some projects, and Colab provides one.

Course Structure

• Unit 1 — Foundations: Learning from Data (Weeks 1–3, 8 lectures): the learning problem, nearest neighbors, the math toolkit, linear regression, generalization, regularization.
• Unit 2 — Classification & Classic Supervised Learning (Weeks 4–7, 11 lectures + project studio + midterm): logistic regression, evaluation, generative classifiers, Bayesian inference, SVMs and kernels, trees and ensembles, learning theory, the practical workflow.
• Unit 3 — Unsupervised Learning & Probabilistic Models (Weeks 8–11, 6 lectures): clustering, mixtures and EM, PCA, embeddings, recommender systems, graphical models.
• Unit 4 — Neural Networks & Modern ML (Weeks 12–16, 13 lectures + presentations): neural networks and backpropagation, training, CNNs, attention and Transformers, self-supervised and foundation models, generative models, reinforcement learning, fairness and deployment, the frontier.

See the class schedule for the lecture-by-lecture calendar with readings.

Assessment

Letter scale: A ≥ 93, A− ≥ 90, B+ ≥ 87, B ≥ 83, B− ≥ 80, C+ ≥ 77, C ≥ 73, C− ≥ 70, D ≥ 60, F < 60.

Midterm format. The full pool of possible exam problems is released Wed Feb 24, one week before the exam, and every exam problem is drawn from that pool. Work through it however you like — study groups and AI assistants included — but the exam itself is closed book: bring the solutions in your head, not on paper. You take it in class as a Canvas quiz in Respondus LockDown Browser on your own laptop (a paper copy is available if you need one); the Mar 1 project studio doubles as the LockDown setup rehearsal.

CSCI 4750 requirements (vs. CSCI 5750). The two listings share lectures, homework deadlines, the midterm, the final project, and the in-class participation questions. They differ in one place:

• Homework: each of HW1–HW4 includes a marked graduate extension (the “G part”). For CSCI 4750 students the G part is optional extra credit (up to +3 points per homework); your homework grade is based on the core parts (out of 100).

Every project team — 4750-only, 5750-only, or mixed — chooses either track: research or application (see the project page).

Policies

Attendance. Lecture is the core of the course and slides are not a substitute. Attendance is not directly graded, but in-class participation feeds the participation grade.

Late work. Each student has 5 free late days for the semester, at most 2 per assignment, applied automatically. Beyond that, −5 points per additional day late (partial days are rounded down). Late days do not apply to the midterm or to the project proposal, presentation, or final report (hard deadlines because they gate feedback and scheduling).

Collaboration. Discussing ideas is encouraged; sharing or copying code or written answers is not. Each submission lists everyone you discussed it with.

AI tools policy. Machine learning is the science behind today’s AI assistants, so using them thoughtfully is part of what you learn here, not a loophole. Three tiers, stated on every assessment:

• Tier A — Required or freely permitted: designated parts (HW4 Part 3, and parts of many projects) ask you to use pretrained models, and AI assistants are permitted freely there, with disclosure.
• Tier B — Permitted with disclosure (default for homework): you may use AI assistants for homework the way you’d use a colleague — to explain concepts, debug, or draft — provided
  1. you disclose the tool and how you used it, and (2) you can explain every line you submit. Instructors may ask you to walk through your solution; inability to explain it is treated as an integrity violation. Core “from scratch” functions flagged ★ in each assignment must be your own typing.
• Tier C — Prohibited: the in-class midterm (closed book). In-class participation questions are answered by you, in the room, without AI tools — they are credited for participation, not correctness, so there is nothing to gain anyway.

The assessment mix (in-class exam + oral project presentation + explainability rule) is deliberately designed so the course measures what students can do, not what their tools can do. (This section also serves as the written generative-AI use policy that SLU requires every course syllabus to state explicitly.)

Academic integrity. The University’s required statement appears in University Syllabus Statements below; suspected violations in this course are handled under the University-level Academic Integrity Policy.

Accommodations, Title IX, wellness. See the required and recommended University statements in University Syllabus Statements below.

University Syllabus Statements (AY26-27)

The statements below are the official SLU institution-level syllabus statements, effective August 2026 – July 2027, taken verbatim from the Provost’s compiled statements document (Title IX statement revised 8/13/26; others unchanged from last year).

Academic Integrity

Academic integrity is the commitment to and demonstration of honest and moral behavior in an academic setting. Since the mission of the University is “the pursuit of truth for the greater glory of God and for the service of humanity,” acts of integrity are essential to its very reason for existence. Thus, the University regards academic integrity as a matter of serious importance. Academic integrity is the foundation of the academic assessment process, which in turn sustains the ability of the University to certify to the outside world the skills and attainments of its graduates. Adhering to the standards of academic integrity allows all members of the University to contribute to a just and equitable learning environment that cultivates moral character and self-respect. The full University-level Academic Integrity Policy can be found on the Provost’s Office website at: https://www.slu.edu/provost/policies/academic-and-course/academic-integrity-policy.pdf.

Disability Accommodations

Students with a documented disability who wish to request academic accommodations must formally register their disability with the University. Once successfully registered, students also must notify their course instructor that they wish to use their approved accommodations in the course.

Please contact the Center for Accessibility and Disability Resources (CADR) to schedule an appointment to discuss accommodation requests and eligibility requirements. Most students on the St. Louis campus will contact CADR, located in the Student Success Center and available by email at accessibility_disability@slu.edu or by phone at 314.977.3484. Once approved, information about a student’s eligibility for academic accommodations will be shared with course instructors by email from CADR and within Banner. Students who do not have a documented disability but who think they may have one also are encouraged to contact CADR. Confidentiality will be observed in all inquiries.

Title IX

Saint Louis University and its faculty are committed to supporting our students and seeking an environment that is free of bias, discrimination, and harassment. If you have encountered any form of discrimination on the basis of sex, including sexual harassment, sexual assault, stalking, domestic or dating violence, we encourage you to report this to the University. If you speak with a faculty member about an incident that involves a Title IX matter, that faculty member must notify SLU’s Title IX Coordinator that you shared an experience relating to Title IX. This is true even if you ask the faculty member not to disclose the incident. The Title IX Coordinator will then be available to assist you in understanding all of your options and in connecting you with all possible resources on and off campus.

If you are pregnant or experiencing a pregnancy related condition, the Title IX Coordinator can assist you in understanding your rights and options as well as provide supportive measures.

Anna Kratky is the Title IX Coordinator at Saint Louis University (DuBourg Hall, room 36; anna.kratky@slu.edu; 314-977-3886). If you wish to speak with a confidential source, you may contact the counselors at the University Counseling Center at 314-977-TALK or make an anonymous report through SLU’s Integrity Hotline by calling 1-877-525-5669 or online at SLU.EDU/INTEGRITYHOTLINE. To view SLU’s policies, and for resources, please visit the following web address: https://www.slu.edu/about/safety/sexual-assault-resources/index.php.

Student Success Center

The Student Success Center (SSC) supports students in reaching their goals in and out of the classroom. Providing a variety of resources, the Student Success Center houses both the Center for Accessibility and Disability Resources (CADR) and Academic Support, which includes Tutoring, Supplemental Instruction, University Writing Services, and Student Success Coaching. The Student Success Center is located in the Busch Student Center, Suite 331, and students can make an appointment with any SSC resource via EAB Navigate - Student. To learn more about the Student Success Center and its resources, please visit: https://www.slu.edu/life-at-slu/student-success-center/index.php.

University Writing Services

University Writing Services offers one-on-one consultations with trained writing consultants who help with everything from brainstorming, outlining, and proposing research questions to documenting sources, revising, and implementing feedback. These consultations are available to both graduate and undergraduate students and can take place in-person, asynchronously, or via Zoom. All appointments can be scheduled through EAB Navigate – Student. Getting feedback benefits writers at all skill levels on different writing projects (including but not limited to class assignments, conference papers, cover letters, dissertations, group projects, multimedia assignments, personal statements, senior capstone projects, short answer questions on applications, speeches, and theses). For additional information, visit https://www.slu.edu/life-at-slu/student-success-center/academic-support/university-writing-services/index.php or send an email to writing@slu.edu.

University Counseling Center

The University Counseling Center (UCC) offers free, short-term, solution-focused counseling to Saint Louis University undergraduate and graduate students. UCC counselors are highly trained clinicians who can assist with a variety of issues, such as adjustment to college life, troubling changes in mood, and chronic psychological conditions. To make an appointment for a wellness consultation, call 314-977-8255 (TALK), visit the UCC website or walk into the clinic on the second floor of Wuller Hall. For after-hours needs, please press #9 after dialing the clinic number and ask for a mental health specialist. Free virtual teletherapy is also available through UCC partner, TimelyCare.

Wellness

With our Jesuit commitment to cura personalis, the University sees your academic success as connected to your health and well-being and provides resources to support your holistic wellness.

All students experience stressors and challenges at some point, and seeking support is both normal and beneficial. Such challenges may be the result of academic concerns (such as those related to particular assignments or content in a course), or they may be more personal in nature (such as concerns related to relationships, mental health, medical issues, loss, identities, alcohol or drugs, housing or food security, finances, or local/world events, among other things). If you experience these or other difficulties that are impacting your well-being and/or academic work, please consider seeking support from the resources available to you.

• For questions or concerns related to this course, please contact me. I am invested in your success and will support your success in the ways I can.
• Additionally, you have access to the many resources SLU provides in support of your personal wellness. You will find a list of available resources on the Well-being page of the SLU website.

If you or someone you know is experiencing a crisis: please consult Crisis Support and Warning Signs on the University Counseling Center website or call the University Counseling Center at 314-977-TALK (8255) and press #9 to be connected to a behavioral health nurse 24/7.

Basic Needs Security

Students experiencing food insecurity, housing insecurity, and any other challenges that are impacting their personal and/or academic wellbeing are encouraged to contact the Dean of Students Office for support. Students can submit an intake form, email deanofstudents@slu.edu, or call 314-977-9378 to connect with their office. Students may also communicate directly with their instructors about any challenges they are experiencing to receive support and resource referrals.

Weekly Cadence

• Lectures: Mon/Wed/Fri, slide-based with worked examples; each lecture closes with a short “Now” segment connecting the day’s method to current industry or research.
• In-class questions: most lectures include a few live multiple-choice questions answered on your own device (scan the QR code shown in class, or open the course website); the answer distribution appears on screen in real time, and answering counts toward the participation grade.
• Homework: released about every three weeks, due Fridays 11:59 PM (see schedule).
• Office hours: Mon/Wed/Fri 11:00 AM – 12:00 PM in Ritter Hall 309; project-focused office hours after spring break.