Page 49 - Ai V2.0 Flipbook C8
P. 49

This way, it can find hidden patterns or trends without anyone telling it exactly what to do. This
                approach is useful because sometimes it’s hard to write clear rules for complicated or messy
                data. The machine’s ability to discover patterns helps solve problems that rule-based systems
                cannot handle easily.

                Let  us understand  the  concept  of a  learning-based AI approach using a  simple example
                of fruit images. In this method, the AI model is given unlabelled data—in this case, various

                fruits  like  apples,  bananas,  grapes,  and  pears—without  any  names  or  predefined  categories.
                The machine does not know what each fruit is called or how it should be grouped. Instead of
                following fixed rules, the AI analyses the visual features of the fruits such as shape, colour, and
                size to find patterns. It then learns from these patterns and clusters similar items together.

                For example, it groups all bananas together based on their long, curved shape and yellow colour,
                and clusters apples together based on their round shape, even if the colours vary. This process
                is known as unsupervised learning, where the machine identifies categories or groups by itself
                without human-provided labels.

                The following image effectively demonstrates how a learning-based AI model can independently
                observe, analyse, and group data based on the features it detects.





                                                                                          Learning Approach


                                                                   Unlabelled Data


                                                                                               Model
                                                                Used to Train Dataset



                                    Unlabelled Data                                                   Output


















                                 Output is clustered based on patterns observed by the machine:
                         Left is based on roller stakes, Middle is based on Ice Skates, and Right is based on
                                                         Inline skates.








                                                                              Stages of AI Project Cycle  47
   44   45   46   47   48   49   50   51   52   53   54