- classification tasks
- supervised learning
- discriminative deep learning models
- artificial neural networks (ANNs)
- deep neural networks (DNNs)
The Deep Learning Bootcamp serves as an extension of the existing 42AI bootcamps: Python & Machine Learning, aiming to introduce participants to advanced concepts in deep learning. Emphasizing consistency with former bootcamps, this program focuses on introducing new notions and concepts specific to deep learning.
Two key considerations include maintaining alignment with past bootcamps and adjusting difficulty levels accordingly.
Given its nature as an extension, the bootcamp may not revisit fundamental concepts already covered in previous bootcamps. Instead, it will delve into more advanced topics across five modules. These modules will include hands-on practices on building neural networks “from scratch” using Python code.
Each module is structured to span one day, although the complexity of the content may necessitate additional time for thorough understanding and application. Adjustments in the duration of each module will be made based on the difficulty of the material covered to ensure comprehensive learning experiences for participants.
This project is a Deep Learning bootcamp WORKING IN PROGRESS by 42 AI.
As notions seen during this bootcamp can be complex, we very strongly advise students to have previously done the following bootcamps:
42 Artificial Intelligence is a student organization of the Paris campus of the school 42. Our purpose is to foster discussion, learning, and interest in the field of artificial intelligence, by organizing various activities such as lectures and workshops.
- sigmoid
- forward
- predict
- binary cross entorpy loss
- deriviatives
- perceptron
- relu
- dense layer
- sgd optimization
- MLP class
- training
- evaluation
- binary classification
Delve into Deep Neural Networks (DNNs), understanding their architecture, training process, and applications.
- tanh
- normalization: z-score, minmax
- momentum optimization
- DNN class
- binary classification
Understand the principles and techniques for classifying data into multiple categories.
- lecky relu
- softmax
- cross entropy loss
- RMSProp optimization
- model architecture
- multiclass classification
Learn to optimize model performance through effective tuning and regularization strategies.
- heyperprameter tuning: grid, random search
- regularization: dropout, l2
- bias-variance trade-off