Symbolic and Neural AI

Studiewijzer 2025/2026

Globale inhoud

Recent neural approaches to artificial intelligence are very effective, with a proliferation of large models trained on correspondingly massive datasets. However, these models still fail on some tasks that humans, and symbolic approaches, can easily solve. There is therefore a need to integrate symbolic and neural approaches, firstly to potentially improve the performance of large neural models, and secondly to analyze and explain the representations that these systems are using. This course will survey key models and approaches that integrate neural, statistical, or distributed approaches  with symbolic, logic-based approaches to AI. In this course you will be expected to read and discuss key academic papers that will be introduced in lectures.

By the end of this course you should be able to:

• describe neural approaches to AI and symbolic approaches to AI
• compare and contrast neural and symbolic approaches
• summarize different approaches to integrating neural and symbolic approaches to AI
• implement (and possibly extend) one or more neurosymbolic models
• identify and discuss real-world applications of neurosymbolic AI

The field of neurosymbolic AI is still new, and definitely an area of active research. A large part of your learning will be through the supported reading of papers in the area. Most weeks, there will be some assigned reading and some programming, and occasionally some pen-and-paper problem sets. You should start the reading early (at the start of the week) and come to the lectures ready to discuss. Often, there will be parts that you don't understand. This is fine, but please prepare questions to ask in the lectures - if you don't understand something then it's likely others won't either!

Studiemateriaal

Literatuur

• Neuro Symbolic Reasoning and Learning, Paulo Shakarian, Chitta Baral, Gerardo I. Simari, Bowen Xi, Lahari Pokala, https://doi.org/10.1007/978-3-031-39179-8. Should be accessible with UvA login, also on Canvas.

• Reading will be made available on Canvas.

Practicummateriaal

• Programming worksheets will be made available on Canvas

Leerdoelen

• Describe neural approaches to AI and symbolic approaches to AI.
• Compare and contrast neural and symbolic approaches.
• Summarize different approaches to integrating neural and symbolic approaches to AI.
• Implement (and possibly extend) one or more neurosymbolic models.
• Identify and discuss real-world applications of neurosymbolic AI.

Onderwijsvormen

• Hoorcollege
• Werkcollege
• Zelfstudie
• Zelfstandig werken aan bijv. project/scriptie

The hoorcolleges will be partly in a 'lecture' format, where the teacher will give an overview of material for the week, and partly in a discussion format. Prior to the hoorcolleges, students are expected to have done the assigned reading, and to come to the hoorcolleges prepared to discuss and ask questions.

The werkcolleges will have some time set aside for an overview of the programming worksheets, and some time for students to work independently on the worksheets and/or reading for that week. There will also be an opportunity for feedback on the previous week's work.

In the self-study time, students are expected to do the weekly reading and complete programming assignments and problem sets. In later weeks, they are expected work towards the final assessment.

Verdeling leeractiviteiten

Aanwezigheid

Aanwezigheidseisen opleiding (OER-B Artikel B-4.10):

• Voor sommige studieonderdelen geldt een aanwezigheidsplicht. Indien er een aanwezigheidsplicht geldt, dan staat dit aangegeven in de studiegids. De onderbouwing voor, en invulling van, deze aanwezigheidsplicht kan per vak verschillen, en is opgenomen in de studiewijzer. Wanneer studenten niet voldoen aan deze aanwezigheidsplicht kan het onderdeel niet met een voldoende worden afgerond.
  

Aanvullende eisen voor dit vak:

Students must attend 10 out of the 12 werkcolleges. Students who do not meet this requirement will not be permitted to resubmit their coursework in the case of failing the course. Students may request exceptions for special personal circumstances as described in the Teaching and Examination Regulations (OER).

Toetsing

Inzage toetsing

On release of grades, students will be invited to request feedback.

Opdrachten

There are 5 assessed and non-assessed assignments. 

Assessed Notebooks: Individually completed, graded out of 10. Written feedback given.

Week 1: Jupyter Notebook and written questions on Logic and Neural Networks

Week 2: Jupyter Notebook on Logic Tensor Networks

Week 3: Jupyter Notebook on Logical Neural Networks

Week 5: Jupyter Notebook on Differentiable Inductive Logic Programming

Each assessed notebook is worth 5% of the total grade, for 20% overall.

Final Project: Completed in pairs, graded out of 10. Written feedback given.

Week 8: Final project and report extending one or more of the assessed notebooks.

Fraude en plagiaat

Dit vak hanteert de algemene 'Fraude- en plagiaatregeling' van de UvA. Hier wordt nauwkeurig op gecontroleerd. Bij verdenking van fraude of plagiaat wordt de examencommissie van de opleiding ingeschakeld. Zie de Fraude- en plagiaatregeling van de UvA: http://student.uva.nl

Weekplanning

Contactinformatie

Coördinator

• Martha Lewis