The goal of any Convolution Neural Network is to learn higher-order features from data representation, achieving that via convolutions. This type of Neural Nets is very good in dealing with tensor data such as images and is well suited to object recognition with consistently top image classification competitions. In this part, I will try to teach you the convolution neural network on weekend.

An online and free version of this part can be found here.

You can check the Scratching Linear Algebra in Weekend here.

Also, do not forget to check the following weekends:

Python Basics

Python OOP

NumPy

Pandas I.

Matplotlib

SciPy

Introduction to Data

Data Visualization and Understanding

Data Cleaning and Preprocessing Techniques.

Data Exploration and Analysis.

Data Exploration Part I.

Data Exploration Part II.

Data Exploration, It's all about Data Wrangling.

The Art of Data Visualization.

Rules of data visualization.

Linear Algebra.

Math for Machine Learning.

Gradient Descent.

Linear Regression.

Logistic Regression.

Support Vector Machines.

Tree Methods.

Ensemble Methods.

Introduction to Deep Learning.

Vanilla Neural Network, A refresher guide.

Convolution Neural Network, A refresher guide.

Recurrent Neural Network, A refresher guide.

Attention Mechanism, A refresher guide.

Auto Encoders, A refresher guide.

Generative Adversarial Networks, A refresher guide.

One of the fields that have been greatly influenced by convolutional neural networks is the automotive industry. Tasks such as pedestrian detection, car detection, traffic sign recognition, traffic light recognition and road scene understanding are rarely done using hand-crafted features anymore.

Designing, implementing and evaluating are crucial steps in developing a successful computer vision-based method. In order to design a neural network, one must have the basic knowledge about the underlying process of neural network and training algorithms. Implementing a neural network requires a deep knowledge about libraries that can be used for this purpose. Moreover, neural network must be evaluated quantitatively and qualitatively before using them in practical applications.