Učni načrt predmeta

Predmet:
Umetna inteligenca in znanost
Course:
Artificial Intelligence for Science
Študijski program in stopnja /
Study programme and level
Študijska smer /
Study field
Letnik /
Academic year
Semester /
Semester
Nanoznanosti in nanotehnologije, Informacijske in komunikacijske tehnologije, Ekotehnologije, Senzorske tehnologije, 3. stopnja / 1 1
Nanosciences and Nanotechnologies, Information and Communication Technologies, Ecotechnologies, Sensor Technologies, 3rd cycle / 1 1
Vrsta predmeta / Course type
Izbirni / Elective
Univerzitetna koda predmeta / University course code:
SPL-906
Predavanja
Lectures
Seminar
Seminar
Vaje
Tutorial
Klinične vaje
work
Druge oblike
študija
Samost. delo
Individ. work
ECTS
20 10 15 105 5

*Navedena porazdelitev ur velja, če je vpisanih vsaj 15 študentov. Drugače se obseg izvedbe kontaktnih ur sorazmerno zmanjša in prenese v samostojno delo. / This distribution of hours is valid if at least 15 students are enrolled. Otherwise the contact hours are linearly reduced and transfered to individual work.

Nosilec predmeta / Course leader:
prof. dr. Sašo Džeroski
Sodelavci / Lecturers:
doc. dr. Panče Panov , prof. dr. Ljupčo Todorovski
Jeziki / Languages:
Predavanja / Lectures:
slovenščina, angleščina / Slovenian, English
Vaje / Tutorial:
Pogoji za vključitev v delo oz. za opravljanje študijskih obveznosti:
Prerequisites:

Izpolnjeni morajo biti pogoji za vpis na doktorski študij: zaključena druga stopnja bolonjskega študija ali diploma univerzitetnega študijskega programa. Potrebna so tudi osnovna znanja računalništva oz. informatike.

Students must fulfill the formal requirements for enrolling to the doctoral study program: completed Bologna second level study program or an equivalent pre-Bologna university study program. Basic knowledge of computer science or informatics is also required.

Vsebina:
Content (Syllabus outline):

Umetna inteligenca (UI) in znanost: Uvod
Osnovni gradniki znanja v znanosti
Principi avtomatizirane znanosti in računalniškega
odkrivanja znanstvenih zakonitosti
Strojno učenje in napovedno modeliranje
Odprta znanost

Formalizmi za predstavitev znanja v znanosti
Od podatkov, preko modelov, do teorij
Znanstvene taksonomije in ontologije
Metode sklepanja za ontologije

Principi odprte znanosti
Reproducibilnost znanstvenih poskusov
Večkratna uporaba raziskovalnih rezultatov
Principi FAIR (najdljivost, dostopnost,
interoperabilnost, večkratna uporabnost)
za znanstvene podatke in modele
Meta-podatki in ontologije za znanost

Strojno učenje za analizo znanstvenih podatkov
Simbolične metode strojnega učenja
Obravnava strukturiranih podatkov
Umetne nevronske mreže in globoko učenje
Obravnava nestrukturiranih in
visokodimenzionalnih podatkov

Primeri uporabe: Študije primerov uporabe metod umetne inteligence in strojnega učenja v znanosti
Uporaba metod UI v fiziki
Primeri iz fizike delcev
Primeri iz znanosti o materialih
Uporaba metod UI v znanostih o življenju
Napovedovanje funkcij genov
Analiza mikrobiomskih podatkov
Virtualno presejalno testiranje spojin

Introduction: Artificial Intelligence (AI) for science
The basic components of scientific knowledge
Principles of automated science and computational scientific discovery
Machine learning (ML) and predictive modeling
Open science

Formal representation of scientific knowledge
From data through models to theories
Scientific taxonomies and ontologies
Reasoning methods for ontologies

Principles of open science
Reproducibility of scientific experiments
Reusability of research outputs
FAIR (findable, accessible, interoperable and
reusable) principles for scientific data and models
Meta-data and ontologies for science

Machine learning for the analysis of scientific data
Symbolic machine learning methods
Handling structured and semi-structured data
Artificial neural networks and deep learning
Handling unstructured and high-dimensional data

Applications: Case studies of using artificial intelligence and machine learning in science
Applications of AI in physics
Particle physics
Materials science
Applications of AI in life sciences
Gene function prediction
Analzying microbiome data
Virtual compound screening

Temeljna literatura in viri / Readings:

Izbrana poglavja iz naslednjih knjig: / Selected chapters from the following books:
• Džeroski S., and Todorovski L., editors. Computational Discovery of Scientific Knowledge: Introduction, Techniques, and Applications in Environmental and Life Sciences. Springer, 2007. ISBN 978-3-540-73919-7.
• S. Džeroski, B. Goethals, and P. Panov, Eds. Inductive Databases and Constraint-Based Data Mining. Springer, 2010. ISBN 978-1-4419-7737-3
• James G., Witten D., Hastie T., and Tibshirani R. An Introduction to Statistical Learning. Springer, 2013. ISBN 978-1-4614-7138-7.
• Arp R., Smith B., and Spear A.D. Building Ontologies with the Basic Formal Ontology. MIT Press, 2015.
• Goodfellow I., Bengio, Y., and Courville, A. Deep learning. MIT Press, 2016. ISBN 978-0262035613.

Cilji in kompetence:
Objectives and competences:

Cilj predmeta je seznaniti študenta s področji umetne inteligence in strojnega učenja, vključno z osnovnimi koncepti in sodobnimi metodami, s poudarkom na njihovi uporabi v znanosti.

Kompetence študenta z uspešno zaključenim predmetom bodo vključevale razumevanje osnovnih pojmov iz področja UI in strojnega učenja, poznavanje sodobnih metod in sposobnost samostojne uporabe teh metod pri reševanju znanstvenih problemov v skladu s principi odprte znanosti.

The course objective is to familiarize the student with the fields of artificial intelligence and machine learning, including basic concepts and state of the art methods, with a focus on their applications in science.

The competencies of the students successfully completing this course will include the understanding of basic concepts from the field, familiarity with the state-of-the art methods, capability of independent use of AI and ML methods for solving scientific problems while following the principles of open science.

Predvideni študijski rezultati:
Intendeded learning outcomes:

• Dobiti pregled obstoječih nalog in metod v umetni inteligenci in strojnem učenju, kot tudi študij primerov njihove uporabe na področju fizike in znanostih o življenju
• Pridobiti sposobnost formulacije konkretnih problemov iz posameznega izbranega znanstvenega področja kot problemov strojnega učenja
• Pridobiti sposobnost ugotavljanja primernosti različnih metodoloških pristopov za reševanje posameznih problemov s strojnim učenjem
• Pridobiti sposobnost slediti splošnim principom odprte znanosti pri lastnem raziskovalnem delu

• Acquiring an overview of existing tasks and methods in artificial intelligence and machine learning and case studies of their use in physics and life sciences
• Obtaining the ability to formulate problems specific to a selected scientific discipline as machine learning problems
• Obtaining the ability to identify the best methodological approach available for solving specific problems with machine learning
• Obtaining the ability to follow the general principles of reproducible science

Metode poučevanja in učenja:
Learning and teaching methods:

Predavanja, vaje, konzultacije, samostojno delo

Lectures, excercises, consultancy, individual work

Načini ocenjevanja:
Delež v % / Weight in %
Assesment:
Ustni izpit
50
Oral Exam
Seminarska naloga
25
Seminar Work
Ustni zagovor seminarske naloge
25
Oral Defense of the Seminar Work
Reference nosilca / Lecturer's references:
1. Simidjievski, N., Tanevski J., Ženko B., Levnajić Z., Todorovski L., Džeroski S. (2018). Decoupling approximation robustly reconstructs directed dynamical networks, New Journal of Physics, 20:113003
2. Kuzmanovski, V., Todorovski, L., Džeroski, S. (2018). Extensive evaluation of the generalized relevance network approach to inferring gene regulatory networks. GigaScience, 7(11):giy118
3. ROY, Bijit, STEPIŠNIK, Tomaž, VENS, Celine, DŽEROSKI, Sašo. Survival analysis with semi-supervised predictive clustering trees. Computers in Biology and Medicine. [Print ed.]. 2022, vol. 141, str. 105001-1-105001-19. ISSN 0010-4825. DOI: 10.1016/j.compbiomed.2021.105001
4. BRESKVAR, Martin, DŽEROSKI, Sašo. Multi-target regression rules with random output selections. IEEE access. 2021, vol. 9, str. 10509-10522. ISSN 2169-3536. DOI: 10.1109/ACCESS.2021.3051185
5. BRENCE, Jure, TODOROVSKI, Ljupčo, DŽEROSKI, Sašo. Probabilistic grammars for equation discovery. Knowledge-based systems. [Print ed.]. 2021, vol. 224, str. 107077-1-107077-12. ISSN 0950-7051. DOI: 10.1016/j.knosys.2021.107077