In this tutorial, we will build a network to predict the class of an image. Therefore we use the *tf_flowers* dataset, which consists of 3,670 and 5 different classes. This tutorial builds upon a tutorial by TensorFlow.

This tutorial follows our well-known workflow:

- Loading the data and preparing our datasets
- Define the model
- Training and evaluating the model

First, let's have a look at our dataset. We are using the *tf_flowers* dataset provided by the TensorFlow Dataset library (*tfds*).

We, therefore, load the dataset in our *get_dataset()* function using *tfds.* …

After we started using just one *linear neuron,* we continue and build our first Artificial Neural Network (ANN). This tutorial is built upon the regression tutorial by Tensorflow.

In this tutorial, we are using a dataset describing cars. We are provided with data as the number of cylinders, horsepower, weight, etc. We need to predict the miles per gallon (MPG). Therefore we use the labeled training data and implement our first ANN.

In the last section of this tutorial are a few ideas that you could try to optimize further.

When starting with Machine Learning and Deep Learning, the number of new vocabularies and concepts can be overwhelming. In this story, we tackle this problem by implementing a *linear neuron* using just NumPy.

A *linear neuron* highly relates to the problem of linear regression. The problem we try to solve is equal: We are applying gradient descent on our regression problem to find good weights ** w**.

In this post, we look at the loss function, gradient descent, and training of our model.

The steps to train a model can be divided into:

- Loading the data and preparing our datasets
- Define…

I am a PhD Student in the field of Machine Learning, living in Germany