91273 - Project Work in Machine Learning

Learning outcomes

At the end of the course, the student is able to apply the knowledge acquired in Machine Learning in order to carry out autonomously a project focusing on a topic agreed upon with the teacher.

Course contents

Three possible tracks:

1. A scientific paper is considered as the base for the development of a new data mining algortihm. The software engineering best practices have to be used to include the project into an open-source framework.
2. An innovative software/tool is chosen for experiments and tests on given datasets.
3. The professor assigns a dataset and a mining task to be developed on it, or the students proposes a dataset that can be accepted by the professor.

If  you follow track 3 do the following:

• visit the Kaggle website;
• choose three competitions of your interest, either active or closed; the requirements are to have not less than some thousands of data units and not lass than two data files of different structure/content type to be considered (train and test data count only once)
• send and email to the teacher with the subject "[Project work] Kaggle competitions choice" and containing the link to the three competition of your interest
• wait for the answer and, when the competition is assigned, do as specified on the "Assessment methods section below"

Readings/Bibliography

Scientific papers and documentation provided by the teacher

Teaching methods

The teacher assigns a project , to be developed individually or by a very small group; tutoring will be available on demand.

Assessment methods

This activity is mainly oriented to develop abilities based on the knowledge acquired in the related course.

After the assignment of the task by the teacher:

1. create a github folder named “surname-projecttitle” and share it with the teacher;
2. create a readme.md describing the project, with objectives, results, pertinent references, and anything else that may be useful for understanding
3. create a sub-folder where software and data will be delivered, or including permanent links to the data if the data is too large
   1. the folder must contain a readme.md with the installation and startup notes, and the prerequisites for operation
   2. the solution should produced as a python notebook with an adequate amount of comments and program outputs, in order to constitute an adequate report of the project development (export the notebook also as a pdf in landscape format); if the software is plain code, instead of a notebook, it is necessary to add a report, written in latex, in landscape format, deliver latex and pdf.
   3. the software must be commented in English, the names of the variables must be meaningful and inspired by English, filenames must start with the project title, possibly a shortcut
4. prepare a powerpoint or latex beamer presentation of 15-20 slides, deliver:
   1. the .pptx or .tex with any complementary files
   2. the pdf
5. agree with the teacher a date for the presentation and possible demonstration of the software
6. Unless otherwise indicated at the time of assignment, the indicative delivery date should be within three months after assignment
   1. at any time it is possible to consult the teacher by e-mail or by requesting an appointment
   2. delays are not a problem, please, at the expiry of the three months, notify via e-mail the need to continue

Teaching tools

• Python Notebooks
• GitHub

Office hours

See the website of Claudio Sartori