Page 209 - Ai_417_V3.0_C9_Flipbook
P. 209

Data Modelling Techniques

                 In AI modelling, we develop different algorithms called models, which can be trained to produce intelligent output.
                 In other words, we write code to make a machine artificially intelligent.
                 AI modelling techniques can be broadly classified into two approaches, namely, rule-based and learning-based.
                 Let us learn about them in detail.


                 Rule-Based Approach

                 Rule-based approach is based on a set of rules and set of facts already fed to the machine to generate the desired
                 output. These models can operate with simple basic information and data. The relationships or patterns in the data
                 is defined by the developer.

                 To explain it further, let's take an example. You have a dataset comprising 100 images of cars and 100 images of
                 cycles. To train your machine, you feed this data and label each image as either a car or a cycle. Now if you test
                 the machine with an image of a car, it will compare with the trained data and according to the labels of the trained
                 data it will identify it as a car. This is called a rule-based approach. The rules given to the machine in this example
                 are the labels assigned to the training data.


                                                                                      Rule-based Approach


                                                                  Labelled Datasets



                                                                Used to Train Machine       Model

















                                                                      Output               Used for Testing  Training Data




                                 Machine Identifies the Image as Car                         Testing Data


                 Learning-Based Approach

                 Learning-based  approach  refers to  a  model where  the  relationships or  patterns in  the  data  are  not  explicitly
                 defined by the developer. In this approach, random data is fed into the machine and the machine develops its
                 pattern or trends based on the data outputs. It is an alternative method to address some of the challenges of rule-
                 based systems.

                 This approach is typically followed when the dataset fed to the machine is unlabelled and too random. The machine
                 tries to extract similar features and clusters them in the same datasets. In the end, the machine tells the trends
                 which are observed in the training data.
                                                                         AI Reflection, Project Cycle and Ethics  207
   204   205   206   207   208   209   210   211   212   213   214