AY 24/25 - Machine Learning and Evolutionary Robotics

This page is about the courses named (actually the same course):

  • Machine Learning (456MI-1) and Evolutionary Robotics (456MI-2), for master program IN23, 6+3 CFUs
  • Introduction to Machine Learning and Evolutionary Robotics (332MI), for master program IN19, 9 CFUs
  • Introduction to Machine Learning (470SM), for master programs SM38, SM36, SM34, SM23, SM28, and SM64, 6 CFUs
  • Introduction to Machine Learning (040CM), for master program SM13, 6 CFUs

Program, goals, requirements # ↰

Detailed program # ↰

Part 1 (48h) # ↰

  • Definition of Machine Learning; examples of applications of ML; taxonomy of ML problems; phases of design, development, and assessment of a ML system; terminology and mathematical notation.
  • Elements of software/programming languages for ML; elements of data visualization.
  • Supervised learning.
    • Supervised learning system assessment: cross-fold validation; accuracy and other metrics; metrics for binary classification (FPR, FNR, EER, AUC) and ROC.
    • Tree-based methods.
      • Decision and regression trees: learning and prediction; role of the parameters and overfitting.
      • Trees aggregation: bagging, Random Forest.
    • Support Vector Machines (SVM).
      • Separating hyperplane: maximal margin classifier; support vectors; learning as an optimization problem; maximal margin classifier limitations.
      • Soft margin classifier: learning, role of the parameter C.
      • Non linearly separable problems; kernel: brief background and main options (linear, polynomial, radial); intuition behind radial kernel; SVM,
      • Multiclass classification with SVM.
    • Naive-Bayes classification.
    • The K-nearest neighbors classifier.
  • Unsupervised learning.
    • Cluster analysis: hierarchical methods, partitional methods (k-means algorithm).
  • Text and natural language applications (text mining).
    • Sentiment analysis; features for text mining; common pre-processing steps.

Part 2 (24h) # ↰

  • Evolutionary Computation (EC).
    • High-level working scheme of an Evolutionary Algorithm (EA); terminology.
    • Generational model; selection criteria; exploration/exploitation trade-off; genetic operators with examples; fitness function; multi-objective optimization and Pareto dominance; debugging of an evolutionary search.
    • EA issues (diversity, variational inheritance, expressiveness); fitness landscape; properties of the representation.
    • Examples of common EAs: GA, GP, GE.
  • Evolutionary Robotics.
    • Brief foundations of Artificial Neural Networks and EC.
      • EA for neuroevolution.
    • Significant examples.
      • Evolution of Soft Robots morphologies (body).
      • Evolution of robotic agents controllers (brain).
      • Simultaneous evolution of body and brain.
    • Choosing the task, the fitness; reality gap.

Goal of the course # ↰

Knowledge and understanding # ↰

  • Know the terminology and common mathematical notation for the key concepts of ML and EC systems.
  • Know and understand the main supervised and unsupervised ML techniques; know the high-level working scheme of EAs.
  • Know and understand the phases of design, development, and assessment of a ML system.
  • Know and understand the main assessment metrics and procedures suitable for supervised and unsupervised ML systems.

Applying knowledge and understanding # ↰

  • Formulate a formal problem statement, using the proper terminology and mathematical notation, for simple practical problems in order to tackle them with ML or EC techniques.
  • Design simple end-to-end ML systems.
  • Experimentally assess simple end-to-end ML systems.

Making judgements # ↰

  • Judge if a problem can be tackled with ML.
  • Judge the technical soundness of a ML system.
  • Judge the technical soundness of the assessment of a ML system.

Communication skills # ↰

  • Describe, both in written and oral form, the motivations behind choices in the design, development, and assessment of a ML system, using the proper terminology and possibly exploiting simple plots.

Learning skills # ↰

  • Retrieve information from scientific publications about ML or EC techniques not explicitly presented in this course.

Requirements # ↰

Basics of statistics: basic graphical tools (i.e., plots) for data exploration; summary measures of variable distribution (mean, variance, quantiles); fundamentals of probability and of univariate and multivariate distribution of random variables; basics of linear regression analysis.

Basics of linear algebra: vectors, matrices, vector and matrix operations.

Basics of programming and data structures: algorithm, data types, loops, recursion, parallel execution, tree.

Familiarity with manipulation of mathematical notation.

Method, language, material # ↰

Language of teaching # ↰

English

Teaching method # ↰

Frontal lectures with slide projection and live annotation; lab activities, under teacher supervision, in dealing with simple problems with ML techniques.

Course material # ↰

Teacher slides and lab sketches # ↰

The course material (teacher’s slides) will be served directly online:

The slides might be updated during the course. Sketches for how to do the lab activities, in the form of R notebooks, are given below; please, fully enjoy the lab activity by not looking at these sketches too early:

The recordings of the lectures will be available on MS Teams.

Suggested textbooks # ↰

  • Gareth James, Daniela Witten, Trevor Hastie and Robert Tibshirani. An Introduction to Statistical Learning, with applications in R. Springer, Berlin: Springer Series in Statistics, 2014. (for the 1st part of the course)
  • Kenneth A. De Jong. Evolutionary computation: a unified approach. MIT press, 2006. (for the 2nd part of the course)

These are just suggestions: a significant part of the course is not based on any specific textbook.

Lectures timetable and course calendar # ↰

The course will start on September, 24th. Lectures will be held in:

  • Room H, building C1, Piazzale Europa Campus on Monday, 16.00-19.00 (in practice, 16.00-18.30)
  • Sala Atti, building C1, Piazzale Europa Campus on Monday, 15.00-18.00 (in practice, 15.00-17.30) (from 25/11/24)
  • Room 3B, building H3, Piazzale Europa Campus on Tuesday, 11.00-13.00 (in practice, 11.00-12.30)
  • Room 2A, building D, Piazzale Europa Campus on Wednesday, 10.00-12.00 (in practice, 10.15-11.45 from 9/10/24; 10.00-11.30 before)

The lectures will be given in person and I recommend being in the room. In compliance with the current regulation students might be required to book a place in the room. The lectures will not be cast in streaming, but the recordings of the lectures will be available on the MS Teams team of the course (code jml3dtr).

Short announcements about lectures schedule # ↰

  • Starting from 9/10/24, the practical time for the lecture of Wednesday will be 10.15-11.45.
  • The lecture of 11/11/24 will be held in room B, building C7 from 16.30 to 19.00 (3 academic hours).
  • On 13/11/24, with usual time and room, ther will be a seminar by prof. Carlos H. G. Ferreira, from Federal University of Ouro Preto, Minas Gerais, Brazil. The abstract is: In this class on Data Representation and Analysis with Graph Models, we will explore the concept of graphs, their types and applications, and how they prove essential in representing complex systems, such as social networks, biology, and computing. We will introduce possible analyses within this model, focusing on structural analysis—which examines the internal organization of a network—and community analysis, which identifies important substructures and patterns. From this, we will see how these analyses can provide deep and practical information for interpreting network-structured data across various fields.
  • The second part of the course, the one on Evolutionary Robotics (just for IN19, IN20, IN23), will start on 18/11/24, with usual times and locations for the lectures.
  • Starting from 25/11/24, the Monday’s lecture will change time and room: new time is 15-18, new room is Sala Atti, building H (close to the current Monday’s room) (see above).

End-of-course test (exam) # ↰

The exam can be done in two ways:

  1. project and written test;
  2. written test only.

In the first case, the final grade is the average of the two grades: the exam is considered failed if at least one of the two grades is <18. In the final overall grade, honors (lode) are awarded if and only if the grade for every part is greater of equal to 30/30 and the average of all the parts exceeds 30/30.

Student must register for the exam session of their interest using the online sistem (esse3). Note that there are deadlines for registration (usually 1 week before the session date).

In any type of content produced by the student for admission to or participation in an exam (projects, reports, exercises, tests), the use of Large Language Model tools (such as ChatGPT and the like) must be explicitly declared. This requirement must be met even in the case of partial use. Regardless of the method of assessment, the teacher reserves the right to further investigate the student’s actual contribution with an oral exam for any type of content produced.

Final project # ↰

The student chooses a problem among a closed, teacher-defined set of problems and proposes a solution based on ML or EC techniques. The expected outcome is a written document (few pages) including: the problem statement; one or more performance indexes able to capture any solution ability to solve the problem; a description of the proposed solution from the algorithmic point of view; the results and a discussion about the experimental assessment of the solution with, if applicable, information about used data. Student are encouraged to form groups for the project: in this case, the document must show, for each student of the group, which activities the student took part in. The project is evaluated according to clarity (≈50%), technical soundness (≈33%), and results (≈17%).

The project assignment is here.

Written test # ↰

It consists of questions on theory and application with short open answers. Each test consists of ≈3 questions with a longer answer and ≈3 questions with a short answer. Some of the questions are on the “theory”: they require to give the definition of a key concept, to tell the differences between a few key concepts, or list the suitable options for some given case. Some of the questions are on the “practice”: they require to describe an algorithm, possibly in the form of pseudocode, sketch a design of an ML system for a given case, solve some simple numerical problem involving quantities. The grade of the written test is the weighted average of the grades obtained for each question, with “long questions” weighting double the “short questions”.