Page 123 - Ai_C10_Flipbook
P. 123

Learning-based Approach








                                                                       Model

                                                                   Model Training




                                 Unlabelled Data                                        Clustering on the Basis of Size

                 A  learning-based  approach  in  AI  allows  the  machine  to  train  on  data  and  adapt  its  model  dynamically.  It
                 modifies itself based on data changes, ensuring adaptability and handling exceptions effectively.

                 For example, a learning-based product recommendation system is an AI model used by e-commerce platforms
                 like Amazon or Flipkart to suggest products to users based on their behaviour, preferences, and purchase
                 history. Unlike rule-based systems that suggest predefined items, this approach learns patterns and adapts to
                 users' interests over time.
                 The system is trained using a large dataset of user interactions, that includes browsing history, purchase history,
                 search queries and reviews/rating. The trained model analyses a user’s current activity in real time and matches it
                 with patterns learned during training. Using machine learning algorithms, the recommendation system identifies
                 shifts in user interests or seasonal preferences. This learning-based recommendation system enables e-commerce
                 platforms to offer a personalised shopping experience, boosting user satisfaction and business revenue.
                 The Learning-based Approach can further be divided into three sections:


                                                          Learning-based Approach




                         Supervised Learning               Unsupervised Learning             Reinforcement Learning


                 Supervised Learning

                 Supervised Learning is a type of machine learning where a model is trained on a labelled dataset. A labelled dataset
                 is the information which is tagged with identifiers of data. Labels are the key component in supervised learning,
                 as they guide the learning process. A label is an information that can be used as the tag for data. For example,
                 students in a class are evaluated based on their performance in exams and assignments. Their performance is
                 categorised into labels such as "Outstanding," "Very Good," "Satisfactory," or "Needs Improvement."
                 Supervised learning is like, how a teacher helps students learn. The teacher provides clear examples and guidance
                 (training data) to teach concepts. Later, the teacher tests the students' understanding with new questions (testing
                 data). Similarly, a supervised learning model uses labelled training data to learn patterns and then applies this
                 knowledge to make predictions on unseen data, improving its accuracy over time.
                 Let’s take an example to understand Supervised Learning:
                 Build a model to predict the type of fruit based on its weight and size.
                 Assume that you have a dataset of fruits with their weights and sizes labelled:
                 •  Apple → Weight: 200 grams, Size: Medium        • Banana → Weight: 120 grams, Size: Long
                 •  Orange → Weight: 150 grams, Size: Round        • Grape → Weight: 5 grams, Size: Small
                                                                          Advanced Concepts of Modeling in AI   121
   118   119   120   121   122   123   124   125   126   127   128