Page 15 - CT_AI_Class-7

Dividing a Dataset

Dividing a dataset is an important step in training AI models. The dataset is split into different
parts to help the AI model learn, improve and then check if it has learned correctly. This process
ensures that the AI can make accurate predictions or decisions based on new data. The dataset
is usually divided into three main parts:

 Training dataset: This part of the data is used to teach the AI models. The models learn from
the patterns and trends in this data, allowing them to understand and recognise important
features.

 Validation dataset: This part of the data is used to evaluate the AI model during the training
process. It helps to make adjustments and improve the model as it is learning.

 Test dataset: This part is used for the final check to see if the AI model has learned correctly.
It is important because the AI model has never seen this data before, so it tests how well the
model generalises to new, unseen data.

Consider an example, you are learning a new chapter in maths, in this:

 The training part is when you learn the new concepts (training dataset).

 The validation part is when you practise solving sums to check your understanding (validation
dataset).

 The test part is when you take the final test to check if you’ve understood everything correctly
(test dataset).

Let us consider another example:
Let’s say you have 100 images of animals. You divide these 100 images into:

 60 images for training

 20 images for validation

 20 images for testing

You train the AI model with the 60 images (training images).
After training, you validate the model with 20 images (validation images).

Finally, you test the model with the remaining 20 images (test images).

Moonwalkers AI shoes are innovative wearable devices
designed by Shift Robotics to help people walk faster. They
use artificial intelligence and motorised wheels to adjust speed
based on walking style. These shoes reduce travel time, improve
efficiency and provide a futuristic commuting experience,
ai lens making everyday movement quicker, smoother and easier.

AI Domains and Applications 13