6.008 Introduction to Inference

Fall 2015

We will use an EdEx site for all handouts and lab submissions; you must be registered to access the course materials.
General Information and Syllabus
Frequently Asked Questions

Instructors: Polina Golland, Lizhong Zheng

This subject covers the analytical and computational fundamentals of extracting information from data. Such technologies are used in

  • machine learning
  • search and retrieval
  • data mining
  • computer vision and imaging
  • voice recognition
  • communication and compression
  • natural language processing
  • robotics and navigation
  • computational biology and bioinformatics
  • medical diagnosis
  • distributed sensing and monitoring
  • finance.

Many of the most successful inference and machine learning algorithms arise out of probabilistic modeling and analysis. If you want to learn the fundamentals of this discipline and see some of what you can do with it, this subject is the place to start.

6.008 provides a solid foundation for more advanced subjects that build on this framework of reasoning. As such, the subject is targeted at (and likely to strongly appeal to) students both across and beyond Course 6 (EECS).

As a new subject in our curriculum, we will engage in considerable experimentation with content and pedagogy. While this is an exciting and important part of course development, it also means that the subject will be rougher and more raw than a subject that has been offered many times. As a result, this offering of 6.008 will especially appeal to students who, as part of taking the subject, relish the role of beta-tester and are eager to contribute to the subject's formative development through their participation.

Course Catalog Description

6.008 Introduction to Inference

Prereq: 6.01
Units: 4-0-8
Lecture: MW10
Recitation: TR1 or TR2

Introduction to probabilistic modeling for problems of inference and machine learning from data, emphasizing analytical and computational aspects. Distributions, marginalization, conditioning, and structure; graphical representations. Belief propagation, decision making, classification, estimation, and prediction. Sampling methods and analysis. Introduction to asymptotic analysis and information measures, and applications.


The official prerequisite is 6.01, which includes a very brief (roughly two-week) introduction to probabilistic reasoning. But students who have had a comparable introduction in some other subject will also be suitably prepared. We emphasize that a more thorough introduction to probability, such as in the form of 6.041, 6.042, 18.05 or 18.440 is not required, as we will develop the necessary foundation in probability as part of 6.008. However, we will assume you are comfortable with (and fluent in) basic mathematics at the level of, e.g., 18.02.

Degree Requirements

The requirements satisfied by 6.008 will change in the future as the department reviews the undergraduate curriculum. The requirements specified in this section are true for the Fall 2015 offering of 6.008 only.

EECS students in 6-2 program can use 6.008 as one of their EE or CS foundation subjects. All EECS students can use 6.008 as one of their math elective or free elective subjects.

All EECS students may petition to take 6.008 instead of 6.042 as one of their math elective subjects and to use it as a prerequisite for more advanced subjects that require 6.042. NOTE: the petition must be filed and approved before the add date in Fall 2015. The goal is to verify that you either have sufficient background in constructing formal proofs (an essential component of 6.042 that more advanced subjects rely on) or that you are committed to acquiring such background on your own. More information can be found in Frequently Asked Questions.

6.008 can also be used by MEng students as one of their restricted elective subjects. As usual, no double counting is allowed.

Header and AUS Subjects

The recently introduced 6.036 Introduction to Machine Learning is a course that can be taken by students interested in pursuing inference and machine learning further, and can be used by students instead of 6.034 to satisfy the CS header requirements. 6.008 is not a prerequisite for 6.036, but the two subjects make a natural sequence for students interested in the topics of inference and machine learning.

We currently offer several graduate subjects in inference and machine learning that can be taken as AUS subjects, including 6.437 Inference and Information, 6.438 Algorithms for Inference, and 6.867 Machine Learning.