In 2011, the book Thinking, Fast and Slow introduced a fundamental concept in human cognition: we operate using two modes of thinking and problem-solving:
- Fast Thinking (System 1): Intuitive, automatic, and subconscious—like recognizing a face or driving a familiar route.
- Slow Thinking (System 2): Deliberate, logical, and effortful—like solving a math problem or planning a strategy.
Modern AI, particularly deep learning, excels at System 1 tasks but struggles with reasoning, abstraction, and decision-making—key aspects of System 2 thinking.
This course, System 2, explores AI techniques beyond deep learning to enable structured reasoning and problem-solving. Topics include:
- Neurosymbolic methods
- Program synthesis
- LLMs & Test-time Computation
- Introduction to RL
- LLMs & RL
- Test-time Training/Continual Learning
- Uncertainty Estimation in LLMs
- LLMs and abstraction
- Diffusion Language Models
- LLM Agents
- Open-endedness and Creativity
- CoT Faithfulness
- Complexity OOD
Please be in touch with us through the Telegram channel at https://t.me/system2_spring2026. For necessary cases, contact the instructors or the head TA via email. Make sure to use a clear subject line and include your name and student ID. To facilitate identification by the instructional team, we strongly recommend using the Sharif webmail service.
- Prof. Rohban (Instructor): rohban@sharif.edu
- Prof. Soleymani (Instructor): soleymani@sharif.edu
- Mr. Samiei (Instructor): mm.samiei@sharif.edu
- Mr. Salimi (Head TA): s.salimi.moein@gmail.com
Essential: Deep Learning
Recommended: Planning and Neurosymbolic AI
Classes are held on Sundays and Tuesdays from 15:00 to 16:30 (UTC+3:30) in Room 201, Department of Computer Engineering, Sharif University of Technology.
Please note that grades will be calculated out of 21.0,
with the respective grades for each section provided below:
There is no midterm exam. The final exam will be held on June 28, 2026, at 8:00 AM (UTC+3:30).
There will be three scheduled quizzes (and one optional unscheduled quiz).
There will be four homework assignments in this course, all to be submitted via the Quera Website. You have an additional two days after the deadline for late submissions. Throughout the semester, you are allowed a total of five days of delayed submissions without penalty. Beyond this limit, a 1% deduction will be applied to the assignment grade for every hour of further delay. Due to the tight schedule, no deadline extensions will be granted during the semester.
Key project milestones (from the course Events): Project Topic Selection deadline May 19, 2026; Proposal Presentation June 3, 2026; Project Progress Presentation to TAs (Part 1) June 11, 2026, (Part 2) July 26, 2026; Project deadline August 1, 2026. Further details will be announced on the Telegram channel.
Cheating or copying homework assignments will not be tolerated. On the first offense, you will be referred to the committee and will fail the course. Please review the Education Committee's guidelines on homework integrity.
| Date | Topic | Instructor | |
|---|---|---|---|
|
:
1 |
Date:
Sun, Esfand 3 |
Topic:
Introduction & Motivation |
Instructor:
Dr. Rohban & Mr. Samiei |
|
:
2 |
Date:
Tue, Esfand 5 |
Topic:
Introduction & Motivation |
Instructor:
Dr. Rohban & Mr. Samiei |
|
:
3 |
Date:
Sun, Esfand 10 |
Topic:
Neuro-symbolic |
Instructor:
Dr. Rohban |
|
:
4 |
Date:
Tue, Esfand 12 |
Topic:
Neuro-symbolic |
Instructor:
Dr. Rohban |
|
:
5 |
Date:
Sun, Esfand 17 |
Topic:
Neuro-symbolic |
Instructor:
Dr. Rohban |
|
:
6 |
Date:
Tue, Esfand 19 |
Topic:
Program Synthesis |
Instructor:
Mr. Samiei |
|
:
7 |
Date:
Sun, Esfand 24 |
Topic:
LLMs & Test-time Computation |
Instructor:
Mr. Samiei |
|
:
8 |
Date:
Tue, Esfand 26 |
Topic:
LLMs & Test-time Computation |
Instructor:
Mr. Samiei |
|
:
9 |
Date:
Sun, Farvardin 16 |
Topic:
LLMs & Test-time Computation |
Instructor:
Dr. Soleymani |
|
:
10 |
Date:
Tue, Farvardin 18 |
Topic:
LLMs & Test-time Computation |
Instructor:
Dr. Soleymani |
|
:
11 |
Date:
Sun, Farvardin 23 |
Topic:
Introduction to RL |
Instructor:
Dr. Rohban |
|
:
12 |
Date:
Tue, Farvardin 30 |
Topic:
Introduction to RL |
Instructor:
Dr. Rohban |
|
:
13 |
Date:
Sun, Ordibehesht 1 |
Topic:
LLMs & RL |
Instructor:
Dr. Soleymani |
|
:
14 |
Date:
Tue, Ordibehesht 6 |
Topic:
LLMs & RL |
Instructor:
Dr. Soleymani |
|
:
15 |
Date:
Sun, Ordibehesht 8 |
Topic:
Test-time Training/Continual Learning |
Instructor:
Dr. Soleymani |
|
:
16 |
Date:
Tue, Ordibehesht 13 |
Topic:
Test-time Training/Continual Learning |
Instructor:
Dr. Soleymani |
|
:
17 |
Date:
Sun, Ordibehesht 15 |
Topic:
Uncertainty Estimation in LLMs |
Instructor:
Dr. Rohban |
|
:
18 |
Date:
Tue, Ordibehesht 20 |
Topic:
Uncertainty Estimation in LLMs |
Instructor:
Dr. Rohban |
|
:
19 |
Date:
Sun, Ordibehesht 22 |
Topic:
LLMs and abstraction |
Instructor:
Dr. Soleymani |
|
:
20 |
Date:
Tue, Ordibehesht 27 |
Topic:
LLMs and abstraction |
Instructor:
Dr. Soleymani |
|
:
21 |
Date:
Sun, Ordibehesht 29 |
Topic:
Diffusion Language Models |
Instructor:
Dr. Rohban |
|
:
22 |
Date:
Tue, Khordad 3 |
Topic:
Diffusion Language Models |
Instructor:
Dr. Rohban |
|
:
23 |
Date:
Sun, Khordad 5 |
Topic:
LLM Agents |
Instructor:
Dr. Rohban |
|
:
24 |
Date:
Tue, Khordad 10 |
Topic:
LLM Agents |
Instructor:
Dr. Rohban |
|
:
25 |
Date:
Sun, Khordad 12 |
Topic:
Open-endedness and Creativity |
Instructor:
Dr. Rohban |
|
:
26 |
Date:
Tue, Khordad 17 |
Topic:
Open-endedness and Creativity |
Instructor:
Dr. Rohban |
|
:
27 |
Date:
Sun, Khordad 19 |
Topic:
CoT Faithfulness |
Instructor:
Dr. Rohban |
|
:
28 |
Date:
Tue, Khordad 24 |
Topic:
Complexity OOD |
Instructor:
Mr. Samiei |