Block Seminar Machine Learning for Natural Language Processing (Fall 2023)

Deep learning is the predominant machine learning paradigm in natural language processing (NLP). This approach not only gave huge performance improvements across a large variety of natural language processing tasks. However, large languages models struggle with rules, logic, reasoning and facts. Therefore this semesters seminar will focus on neuro-explicit models that is neural networks that also incorporate explicit knowledge like logic, math or laws of nature.

Lecturer:  Dietrich Klakow, Vagrant Gautam

Location:  t.d.b

Time: block course in the fall break 2023 however preparations start earlier. Here the specific time line

Closing topic doodle: May 1st
Kick-Off: some time May 8-May 12 (doodle)
One page outline: due June 19
Draft presentation: due July 16

Practice talks and final talks will be during the summer break. Time/date will be decided during the kick-off.

Application for participation:  see CS seminar system (for CS; DSAI, VC, ES, …), for CoLI,LST, LCT use this registration system.

HISPOS registration deadline: July 16

Grading (tentative):

  • 5% one page talk outline
  • 10% draft presentation
  • 10% practice talk
  • 35% final talk
  • 5% contributions to discussion during final talk of fellow participants
  • 35% report

List of Topics (tentative):

Neuro-Symbolic Speech Understanding in Aircraft Maintenance Metaverse
Semantic Probabilistic Layers for Neuro-Symbolic Learning
Neuro Symbolic Continual Learning: Knowledge, Reasoning Shortcuts and Concept Rehearsal
Neurosymbolic AI: The 3гd Wave
Deep probabilistic logic: A unifying framework for indirect supervision
A Semantic Framework for Neural-Symbolic Computing
Symbolic, Distributed, and Distributional Representations for Natural Language Processing in the Era of Deep Learning: A Survey
Generating Instructions at Different Levels of Abstraction
GreaseLM: Graph REASoning Enhanced Language Models for Question Answering
Polygrammar: Grammar for Digital Polymer Representation and Generation
Neuro-symbolic XAI for Computational Drug Repurposing
ExtEnD: Extractive Entity Disambiguation
Reducing Disambiguation Biases in NMT by Leveraging Explicit Word Sense Information
Concept Bottleneck Models
Drug Discovery With XAI Using Deep Learning
Grammar-Induced Geometry for Data-Efficient Molecular Property Prediction
Neuro-symbolic partial differential equation solver