Devoir Surveillé

Langue

: 06/01/2022

: 1h

: Anglais

NB : Explain your work, including showing your choices and preferences. We’ll give partial credit for

good explanations of what you were trying to do. Partial credit for coding style will be given.

All code should be done on Colab. Datasets can be downloaded from the drive through the given links.

Pipeline:

1. Read and Load the Dataset

2. Exploratory Data Analysis

3. Data Visualization of Target Variables

4. Data Preprocessing and Vectorization (tokenization)

5. Splitting our data into Train and Test Subset

6. Model Building

7. Model training and evaluation

8. Model testing

Question 1 (Convolutional neural networks - 10 pts):

Dataset: link

Task: Image classification

Question: Follow the steps involved in the deep learning pipeline to build, train and test a CNN

for animal classification (5 classes) from images according to the given dataset.

Question 2 (Recurrent neural networks - 10 pts):

Dataset: link

Task: Text classification

Question: Follow the steps involved in the deep learning pipeline to build, train and test a RNN-

LSTM for text classification (spam - not spam) according to the given dataset.

1