Advanced concepts of Modelling in AI

Rule-Based Approach (AI Modelling)

An AI approach relying on developer-written logic and explicit IF-THEN rules without autonomous learning from data.

Learning-Based Approach (AI Modelling)

An AI approach where models analyze data to autonomously detect patterns and derive rules for decision-making.

Supervised Learning Framework

A machine learning paradigm trained on labeled datasets with explicit input-output pairs to learn predictive relationships.

Regression Model Task

A supervised learning process designed to predict continuous numeric outcomes (e.g., predicting salary or temperature).

Classification Model Task

A supervised learning process designed to assign data into discrete categorical labels (e.g., detecting spam vs. non-spam).

Unsupervised Learning Framework

A machine learning paradigm that identifies hidden patterns and structures in unlabeled datasets without target answers.

Clustering Technique

An unsupervised learning method that groups data points with shared features into distinct clusters.

Dimensionality Reduction Technique

An unsupervised learning method that decreases input variables while preserving key dataset structure and variance.

Reinforcement Learning Mechanism

A paradigm where an agent learns decision-making policies through continuous environment feedback using rewards and penalties.

Decision Tree Components

A hierarchical model comprising a root node, internal feature testing nodes, decision branches, and leaf output nodes.

Artificial Neural Network (ANN) Architecture

A deep learning structure composed of an input layer, one or more hidden processing layers with weighted connections, and an output layer.

Dataset Splitting Ratio (Train vs. Test)

Partitioning data into a training set ($70\%$–$80\%$) for fitting parameters and a testing set ($20\%$–$30\%$) for evaluating unseen accuracy.

Confusion Matrix Structure ($2 \times 2$)

An evaluation grid summarizing classifier predictions using True Positives ($\text{TP}$), True Negatives ($\text{TN}$), False Positives ($\text{FP}$), and False Negatives ($\text{FN}$).

Accuracy Formula (AI Evaluation)

Calculates total correct predictions relative to all samples: $\text{Accuracy} = \frac{\text{TP} + \text{TN}}{\text{TP} + \text{TN} + \text{FP} + \text{FN}}$

Precision Formula (AI Evaluation)

Calculates the ratio of true positive outcomes to total predicted positive outcomes: $\text{Precision} = \frac{\text{TP}}{\text{TP} + \text{FP}}$

Recall Formula (AI Evaluation)

Calculates the ratio of true positive outcomes to actual positive samples: $\text{Recall} = \frac{\text{TP}}{\text{TP} + \text{FN}}$

F1 Score Formula (AI Evaluation)

Combines Precision and Recall into a single harmonic mean metric: $\text{F1 Score} = 2 \times \frac{\text{Precision} \times \text{Recall}}{\text{Precision} + \text{Recall}}$