Page 178 - AI Computer 10

Step 3: Assign labels to the conversational data, which usually consists of two parts:

● Input (User Query): The actual text or question from the user.
● Output (Bot Response): The response that the chatbot should provide.
Step 4: Divide the labeled dataset into training, validation, and test sets to build and evaluate the model.
Step 5: Using the training dataset, fit the model to learn to associate user queries with their corresponding
responses.

Step 6: Test the model with the test dataset to assess its accuracy and effectiveness in predicting correct
responses.
Step 7: Integrate the trained model into a chatbot framework.

Labelled Data in Machine Learning

Training
Data

Training Prediction

Annotation
circle triangle square

triangle circle square
Labelled Data

Unsupervised Learning

Unsupervised Learning is a type of machine learning that involves receiving a training dataset that is without
any labels. The machine is allowed to discover patterns and insights without any explicit guidance or instruction
from any human expert. Let us understand the concept of an unsupervised learning model with the help of a
simple example.

Consider a dataset of 1000 images of books. You
want to understand some pattern out of the data. To
do this, you can feed the data about the books into
the unsupervised learning model. The model goes
through the data and identifies common features,
such as name of author, name of publisher, genre
of the book, price, etc. and creates groups based on
the feature. One data value may appear in several
groups. The model might come up with features
which are already known to the user, or some
feature which was not considered so far, such as the typeface used on the cover of the book.

Another example of clustering model is customer segmentation on online shopping websites based on their
shopping behaviour, spending habits, and purchasing frequency.

The unsupervised learning model is used for:
u finding unknown patterns in data.

u finding features which can be useful for categorisation.
44
44