Human language processing:
new speech and text processing techniques
Goals
The main objectives of the module are to understand the complexity of human language analysis and computational linguistics, explore the potential of language models based on artificial intelligence, analyze how language models are converging with knowledge representation techniques and analysis semantic, present the latest techniques and advances in speech and natural language processing and provide practical examples where these techniques can be applied in real case studies.
Program
This module belongs to the Advanced training program of Aragón EDIH.
1. Exploring linguistic complexity: computational linguistics, human language and NLP
1.1 Introduction to computational linguistics
1.2 Complexity of analysis and challenges
1.3 Examples
2. The potential of artificial intelligence language models
2.1 The emergence of Transformers and Large Language Models (LLM)
2.2 Advantages and Limitations
2.3 Examples
3. Convergence of two universes: language models (LM) and the representation of knowledge / semantics
3.1. Automatic knowledge extraction based on language models
3.2 Importance of quality. Challenges
3.3 Examples
4. Introduction to speech technologies: use of open source tools for the analysis of multimedia documents
4.1 Fundamentals of spoken language processing
4.2. Deep Learning Tools and Models applied to:
- Speech to text transcription,
- Speaker separation
- Speaker Identification
- Speech generation
4.3. Workshop on the use of open source tools for processing audio contained in multimedia documents
5. Introduction to multimodal language models
5.1. Fundamentals of multimodal language models
5.2 Multimodal language models and tools
5.3 Application development workshop with OpenAI CLIP, BLIP, …
Who is this course aimed at?
The course would be aimed at anyone interested in acquiring advanced skills and knowledge in the field of Speech and Natural Language Processing (NLP) and semantic analysis, and who wishes to explore the latest techniques and applications of artificial intelligence in this constantly evolving field. evolution.
Requirements
- Fundamentals of Artificial Intelligence: Understand the concepts of Machine Learning, mainly supervised and unsupervised learning. Useful to know Reinforcement Learning concepts.
- Fundamentals of Deep Learning applied to language: Basic concepts of deep neural networks and their application to the field of language: feed-forward, LSTM, Transformers.
- NLP (Natural Language Processing): Know the fundamental ideas about natural language processing; task types: audio, text, classification, information retrieval, chatbots; and most common problems: normalization, ambiguity, languages, reasoning, etc.
- Fundamentals of knowledge graphs: Basic knowledge of knowledge graphs and information structuring: RDF, ontology, ...
- Python and Deep Learning Libraries: Basic knowledge of Python programming and familiarity with popular Deep Learning libraries such as Scikit-Learn, Transformers.
- Gmail account, familiarization with the use of Notebooks/Google Colab
- Laptop
Modality
Face to face
Faculty
- Rafael del Hoyo
Big Data and Cognitive Systems Team of the Technological Institute of Aragon
- Paula Peña
Big Data and Cognitive Systems Team of the Technological Institute of Aragon
- Rosa Montanés
Big Data and Cognitive Systems Team of the Technological Institute of Aragon
- Eduardo Lleida
University Professor. Vivolab, Engineering Research Institute of Aragon, I3A, University of Zaragoza
- Antonio Miguel
Hired Professor Doctor. Vivolab, Engineering Research Institute of Aragon, I3A, University of Zaragoza
- Alfonso Ortega
University Professor. Vivolab, Engineering Research Institute of Aragon, I3A, University of Zaragoza
Horario, fecha y lugar
- Total duration: 18 hours
- Dates: October 1, 2, 3, 8, 9 and 10
- Hours: 4:00 p.m. – 7:00 p.m.
- Place: Technological Institute of Aragon. C/ María de Luna, 7 (white building). 50018 Zaragoza
- Maximum number of attendees: 15 people
