Description of the course

The course gives 6 CFU in the Master degrees in Artificial Intelligence and Robotic (MARR)

Objectives

The objectives of this course are to present a wide spectrum of Machine Learning methods and algorithms, discuss their properties, convergence criteria and applicability. The course will also present examples of successful application of Machine Learning algorithms in different application scenarios.

Syllabus

• Classification

  • Basic concepts and evaluation

  • Decision Trees

  • Bayes Learning

  • Linear Models

  • Support Vector Machines

  • Kernels

  • Multiple classifiers

• Regression

  • Linear and logistic regression

  • Instance based (K-NN)

  • Perceptron

  • Neural networks

  • Deep neural networks (CNN)

• Unsupervised learning

  • Clustering (k-Means)

  • Latent variables (EM)

• Reinforcement learning

  • MDP 

  • Q-learning

Teaching Material

• Provided in the Google Classroom system

IMPORTANT  - Read before registering to Classroom -   Registration with classroom system will reveal to all the subscribers (i.e., teachers and other students) your name and photo used in your Google account. Please make sure the account you use to register to classroom contains relevant, appropriate and privacy-preserving information (e.g., just your first and last name or matricola code), as they will be used also for the evaluation.

• Classroom registration code:   ezlharo
  
  

Other Suggested textbooks

• • Machine Learning, Tom Mitchell.

  • Pattern Recognition and Machine Learning, Chris Bishop

  • Machine Learning: a Probabilistic Perspective, Kevin Murphy

  • Deep Learning, Ian Goodfellow, Yoshua Bengio, Aaron Courville

Exams

The main validation method will be a written test during the regular exam sessions. (For very old exam of 9 CFU, the exam contains an additional part on Reinforcement Learning.)

During the lectures, teacher will assign some homeworks.

The homeworks are not mandatory and if delivered within the specified deadlines will be used as bonus marks for the final grade.

To attend an exam, it is mandatory to:

• • register to the exam with the system INFOSTUD

  • bring an ID card

Homeworks

Two homeworks will be assigned during the course. Each homework consists in the solution of a learning problem.

Each student has to submit an individual report (PDF file) and the source code (ZIP file without data) through the Google Classroom assignment (e-mail submissions will not be accepted). Please notice that the report (PDF file) must not be included in the ZIP file.

Only individual reports will be accepted. In case of collaborative work, each student has to submit an individual report in which s/he clearly specifies which parts have been developed jointly and which is his/her individual contribution. Make sure your report contains your name and matricola code.

The technical report should contain: a precise description of the learning problem considered, a description of the algorithm used (including parameters and other implementation choices), and a description of the results obtained (with details about the evaluation procedure). The report should also compare at least two different scenarios (e.g., different algorithms, different parameters of an algorithm, different data sets, etc.).

The source code should contain the code actually developed by the student (excluding external libraries and data sets). In case of use of other software (e.g., examples found on Internet), clearly specify which portion have been modified by the student. In case of joint work with other students, clearly specify the portion of code developed by each single student.

Each homework delivered within the deadline and describing the work done in a good quality report will grant up to 2 extra points in the final grade of the exam. A maximum of 4 extra points can be gained with homeworks.

Homework cannot be delivered after the deadlines (no motivation is acceptable).

Homework bonus points will remain valid for all the academic year in which they are developed and can be summed only to an exam given in the same academic year.

Exam dates

Exam dates are available in the web site of the degree programme.

A.Y. 2024/25

https://corsidilaurea.uniroma1.it/en/corso/2024/30431/programmazione 

A.Y. 2023/24

https://corsidilaurea.uniroma1.it/en/corso/2023/30431/programmazione 

A.Y. 2022/23

https://corsidilaurea.uniroma1.it/en/corso/2022/30431/programmazione

NOTE: March/April and October/November sessions are reserved to part-time and out-of-course students

Infostud registration is mandatory to attend the exams. Students registered to an exam will receive more information from the teacher by email. In case of conflicting information, the e-mail received from the teacher a few days before the exam should be considered.