| Kursusnavn (dansk): | Natural Language Processing and Deep Learning |
| Kursusnavn (engelsk): | Natural Language Processing and Deep Learning |
| Semester: | Efterår 2018 |
| Udbydes under: | cand.it., softwareudvikling og -teknologi (sdt) |
| Omfang i ECTS: | 7,50 |
| Kursussprog: | Engelsk |
| Kursushjemmeside: | https://learnit.itu.dk |
| Min. antal deltagere: | 15 |
| Forventet antal deltagere: | 0 |
| Maks. antal deltagere: | 70 |
| Formelle forudsætninger: | Before the course, the student should:
• be able to use basic algorithms and data structures when programming (equivalently, have passed the Algorithms and Data Structures course)
• be able to use basic algorithmic techniques to design algorithms for a given problem
• be able to carry out basic n-dimensional matrix/vector computations by hand (e.g., matrix and vector products, addition, scalar multiplication, subtraction, etc.)
• hold a deep understanding of basic concepts in Statistics (e.g., distributions, summary statistics, law of large numbers, …)
• be a strong Python programmer, but with some understanding of statically typed, pre-compiled languages (like C++ or Java). |
| Læringsmål: | After the course, the student will be able to:
• Describe three core Natural Language Processing (NLP) tasks and implement basic respective computational approaches: language modelling, POS tagging, syntactic parsing.
• Identify and formulate a task for NLP.
• Identify why a given Neural Network architecture may be appropriate for an NLP task.
• Design and carry out a sound experimental method for Neural-Network based NLP research.
• Analyse the results of an NLP experiment.
• Find, extract and explain results in the NLP and Deep Learning research literature relevant for a given problem. |
| Fagligt indhold: | • An introduction to the major core topics in Natural Language Processing: language modelling, POS tagging and syntactic parsing.
• An introduction to basic Neural Networks for Natural Language Processing, and a corresponding Python-based framework.
• A project-based introduction to current Deep-Learning-based Natural Language Processing research. |
| Læringsaktiviteter: | 12 forelæsninger og 12 øvelsesgange The course will be divided into two phases (12 weeks total):
1) Lecture phase: A 6 week lecturing phase in which student are lectured and carry out appropriate exercises. This phase is meant to introduce the core NLP tasks and basic learning with Neural Networks.
2) Research phase: A 6 week project phase involving the re-implementation of a currently state-of-the-art-competitive Deep Learning-based system from Natural Language Processing research. |
| Obligatoriske aktivititer: | • During the Lecture Phase students hand in a small selection of written or coding exercises each week (at least 5).
• During the Research Phase students provide at least 5 weekly 10-15 minute oral presentations covering a paper related to their project, and an update on their projects, for feedback from the class and lecturers. |
| Eksamensform og -beskrivelse: | D11: Aflevering med mundtlig eksamen baseret på aflevering., (7-scale, internal exam) The exam consists of three parts:
1) An oral exam.
2) A written report.
3) Project code.
Duration of the oral exam: 20 minutes per student.
|
| | |
Kursusoversigt