Page 140 - AI Computer 10

The steps involved in an AI project life cycle are as follows:

u Problem Scoping: The first step of an AI project life cycle is defining the scope of a problem. By scoping a
problem, we are able to develop a working model of how things are. In this step, nature, complexity level
and boundaries of a problem are defined using 4W’s framework— Who, What, Where, and Why.
u Data Acquisition: The process of identifying and gathering all the data requirements for an AI project is
called Data Acquisition. Data Acquisition plays an important role because the whole project is carried out
on the basis of identified requirements.

 Understanding the problem
Problem  4Ws Canvas
Scoping
 Gathering relevant data
Data  Surveys, web scrapings, sensors, APIs
Acquistion
 Visualising and interpreting data
Data  Graphical and Visual tools
Exploration
 Developing AI model
Modelling  Rule-based and Learning-based models
 Testing efficiency of model
Evalauation  Precision, Accuracy, Recall, F1 score
 Integrating AI model into existing systems
Deployment  Cloud-based, on-premises, web apps

u Data Exploration: Data Exploration, one of the most important phases of the AI project cycle, is the process
of understanding the nature of data in terms of quality, characteristics, etc. that you have to work with.
Good quality data is a must for an effective AI world.
u Modelling: In the Modelling phase, collected data must be analysed according to the gathered project
requirements. After analysis, we can train the model using appropriate machine-learning algorithms on the
basis of selected datasets.

u Evaluation: The Evaluation stage involves testing the created model on different datasets to measure its
efficiency across the required parameters. This stage helps in rectifying any errors or bottlenecks in the
model and improve it.

u Deployment: Once the AI model has been evaluated and found to be satisfactory, it is integrated in real-
world situations to ensure successful operation. This ensures value and satisfaction to all the users and
stakeholders involved in the AI project.
Let us understand the AI Project Cycle with the help of an example.

Example: Invitees for Annual Function
u Problem Scoping: The problem involves identifying all the participants, collecting their relevant characteristics
and implementing a security system to prevent any unauthorised persons from attending the function.
u Data Acquisition: The problem needs to gather list of all guests and participants, along with their
characteristics, such as addresses, unique identification numbers, contact numbers, biometrics, etc.
u Data Exploration: The collected data is studied to extract the relevant information which can be implemented
to generate unique identification codes or metrics to differentiate between authorised and unauthorised
persons. The project engineers should also take care to ensure the authenticity of the collected data.
u Modelling: After the relevant data has been collected, several models based on different algorithms can
be devised. The best model can be selected out of the alternatives and developed using the available
technologies. The developed model can be trained using selected datasets that fulfil all project requirements.

6 6