This training provides theoretical and practical knowledge of Natural Language Processing (NLP). It covers the main concepts, from classical statistical techniques to state-of-the-art deep learning models. The training includes practical application using Python to acquire operational skills.
Anyone wishing to acquire theoretical and practical knowledge of NLP with Python.
Prerequisites:
Knowledge of Python.
Basics of statistics and/or machine learning.
Understand the opportunities and use cases related to textual data.
Acquire the fundamental concepts of NLP, including the most advanced techniques.
Be able to implement a business project involving textual data.
Introduction to NLP
- History and evolution of NLP
- Specificities of textual data
- Preprocessing of textual data
- Traditional NLP tasks (information extraction, named entity recognition, etc.)
NLP with statistical techniques
- TF-IDF technique
- Topic detection and LDA algorithm
- Practical application with Python
NLP with machine learning
- Main machine learning algorithms
- Overfitting and regularization of ML algorithms
- Training ML algorithms on textual data
- Practical application with Python
Distributed word representations
- Limitations of one-hot representations
- Co-occurrence matrices
- CBOW and skip-gram
- Classic models (Word2Vec, GloVe, fastText)
- Practical application with Python
NLP with deep learning.
- Fundamentals of deep learning
- Multilayer perceptron
- RNN and LSTM models
- Seq-to-Seq models
- Practical application with Python
Self-supervised and multimodal learning
- Attention mechanism and Transformers
- Fine-tuning (BERT)
- Few-shot learning (GPT-3)
- Multimodal learning
- Practical application with Python
NLP project
The training includes a theoretical course (adapted to the audience’s background) and a practical application with Python for each technique covered.
Business use cases.
Issuance of a certificate of attendance.
Address: Color Business Center, 19 rue de l’industrie L-6089 Bertrange
For in-house training, please contact us.
