OVERVIEW

The teaching unit introduces the key concepts associated with deep learning and exemplifies its applications in the fields of Earth observation from space and of autonomous systems.

AIMS AND CONTENT

LEARNING OUTCOMES

Upon successful completion of this teaching unit, students will be able to analyze, evaluate, and develop deep learning models and algorithms applied to Earth observation from space and to autonomous systems, with particular attention to task such as anomaly detection, planning and decision making, image semantic segmentation, and generative modeling. Students will be provided with methodological basis as well as with practical expertise within common frameworks for deep learning such as TensorFlow or PyTorch.

AIMS AND LEARNING OUTCOMES

After the teaching unit, the student shall know basic notions about neural networks, deep learning, convolutional neural networks, recurrent neural networks, transformers, and generative networks. The students shall also know how to apply deep neural models to satellite imagery and to autonomous systems.

PREREQUISITES

There are no specific requirements, in addition to the normal bases of mathematics and physics that the students are supposed to have from their B.Sc. backgrounds in engineering. Highly recommended proficiency in the Python course.

TEACHING METHODS

Class lectures and software laboratory exercises.
Students with a certified learning disability (DSA), a disability, or other special educational needs are invited to contact the instructor at the beginning of the course to discuss teaching and examination arrangements that, while respecting the learning objectives of the course, take individual learning needs into account and provide appropriate accommodations.
Please also note that requests for exam accommodations or exemptions must be submitted using the form available at https://modulionline.unige.it/richiesta-adattamenti#no-back, to the course teacher, the DITEN contact person (silvana.dellepiane@unige.it), and the relevant office (inclusione.studenti@info.unige.it) at least seven working days before the examination, in accordance with the guidelines available at https://unige.it/disabilita-dsa/richiesta-servizi.

SYLLABUS/CONTENT

Neural networks, multilayer perceptrons, training, regularization
Convolutional neural networks
Recurrent neural networks
Transformers
Common deep learning architectures (encoder-decoder, generative adversarial, etc.)
Generative AI
Applications to Earth observation from space (remote sensing imagery)
Applications to autonomous systems

RECOMMENDED READING/BIBLIOGRAPHY

Class slides will be provided to the students through AulaWeb.
Bishop C., Bishop H., Deep learning, Springer, 2024
Goodfellow I., Bengio Y., Courville A., Deep learning, MIT Press, 2016
R. Cresson, Deep learning for remote sensing images with open source software, CRC Press, 2020.

TEACHERS AND EXAM BOARD

LESSONS

LESSONS START

https://corsi.unige.it/en/corsi/11962/studenti-orario

Class schedule

The timetable for this course is available here: Portale EasyAcademy

EXAMS

EXAM DESCRIPTION

Mandatory written examination about the topics in the syllabus of the teaching unit, with maximum admissible mark equal to 24/30. If a student obtains a sufficient mark in this written exam, then he/she can also optionally take an additional oral examination with maximum admissible mark equal to 30/30 with honors.

ASSESSMENT METHODS

Within the mandatory written examination, the student's knowledge of the main concepts discussed in the teaching unit shall be evaluated. Within the optional oral examination, the student's capability to address simple problems of deep learning form space and autonomous systems and his/her capacity to critically discuss the related methodological bases shall be assessed.

FURTHER INFORMATION

Ask the professor for other information not included in the teaching schedule.