Understanding the Basics of Machine Learning: Essential Concepts Explained

Understanding the Basics of Machine Learning: Essential Concepts Explained

Sharing the key concepts of machine learning or someone else can call the keyword of machine learning dictionary. Let’s learn all the keywords, and then we will break down each point in depth. Here are the all key concepts.

  1. Artificial Intelligence (AI)

  2. Machine Learning

  3. Algorithm

  4. Data

  5. Model

  6. Model fitting

  7. Training Data

  8. Test Data

  9. Supervised Learning

  10. Unsupervised Learning

  11. Reinforcement Learning

  12. Feature (Input, Independent Variable, Predictor)

  13. Feature engineering

  14. Feature Scaling (Normalization, Standardization)

  15. Dimensionality

  16. Target (Output, Label, Dependent Variable)

  17. Instance (Example, Observation, Sample)

  18. Label (class, target value)

  19. Model complexity

  20. Bias & Variance

  21. Noise

  22. Overfitting & Underfitting

  23. Validation & Cross Validation

  24. Regularization

  25. Batch, Epoch, Iteration

  26. Parameter

  27. Hyperparameter

  28. Cost Function (Loss Function, Objective Function)

  29. Gradient Descent

  30. Learning Rate

  31. Evaluation

Artificial Intelligence (AI)

It is the field of computer science focused on creating systems or machines capable of performing tasks that typically require human intelligence. These tasks include problem-solving, learning, decision-making, natural language understanding, and pattern recognition.

AI is divided into 3 types and these are :

  1. Narrow AI (Weak AI)

  2. General AI (Strong AI)

  3. Super AI

Machine Learning

Machine Learning is a subset of Artificial Intelligence that focuses on algorithms and models that learn patterns from data and improve their performance without being explicitly programmed. It’s broadly categorized into:

  1. Supervised Learning

  2. Unsupervised Learning

  3. Reinforcement Learning

Algorithms

A Machine Learning (ML) algorithm is a set of instructions or mathematical models that allow computers to learn patterns from data, make predictions, and improve over time.

Examples of Supervised Learning Algorithms:

  1. Linear Regression

  2. Logistic Regression

  3. Ridge, Lasso, and Elastic Net Algorithms

  4. Support Vector Machines

  5. Naïve Bayes theorem

  6. K-Nearest Neighbors (KNN) Algorithm

  7. Decision Tree

  8. Random Forest

  9. Adaboost

  10. Gradient Boosting

  11. xgboost

Examples of Unsupervised Learning Algorithms:

  1. PCA (Principal Component Analysis)

  2. K Means Clustering

  3. Hierarichal Clustering

  4. DBSCAN Algorithm

Data

Data is a collection of raw materials. Data used in ML is Structured and Unstructured Data.

Model

An ML model is the output of a training process, representing learned patterns in the data.

Model Fitting

Model fitting in machine learning refers to the process of training a machine learning algorithm on a dataset to learn the underlying patterns or relationships between the input features (independent variables) and the target variable (dependent variable)

Training Data

The subset of data used to train the ML model. It includes inputs (features) and their corresponding outputs (labels).

Test Data

A separate dataset is used to evaluate the model's performance on unseen data.

Supervised Learning

A type of ML where models are trained on labeled data, i.e., data with known input-output pairs.

Unsupervised Learning

A type of ML where models learn patterns or structures in data without labeled outputs.

Reinforcement Learning

A learning method where an agent learns to make decisions by interacting with an environment to maximize cumulative rewards.

Feature (Input, Independent Variable, Predictor)

A measurable property or characteristic of the data used as input for a model.

Feature Engineering

The process of transforming raw data into meaningful features to improve model performance.

Feature Scaling (Normalization, Standardization)

Adjusting the range of feature values to bring them to a similar scale.

  • Normalization: Scales data to a range of [0, 1].

  • Standardization: Centers data around the mean with unit variance.

Dimensionality

The number of features (variables) in a dataset.

Target (Output, Label, Dependent Variable)

The variable a model is trained to predict in supervised learning.

Instance (Example, Observation, Sample)

A single data point in a dataset.

Label (Class, Target Value)

The ground truth or actual output value is associated with an instance in supervised learning.

Model Complexity

The capacity of a model to capture patterns. High complexity can lead to overfitting, while low complexity may lead to underfitting.

Bias & Variance

  • Bias: Error due to overly simplistic models (underfitting).

  • Variance: Error due to model sensitivity to small fluctuations in training data (overfitting).

Noise

Irrelevant or random variations in data that don’t represent true patterns.

Overfitting & Underfitting

  • Overfitting: When a model learns patterns specific to the training data, performing poorly on new data.

  • Underfitting: When a model fails to learn the patterns in the data.

Validation & Cross-Validation

  • Validation: Process of assessing model performance on a validation set.

  • Cross-Validation: Divides the dataset into folds to train and test the model multiple times for robust evaluation

Regularization

Techniques (e.g., L1, L2) that constrain model complexity to reduce overfitting.

Batch, Epoch, Iteration

  • Batch: Subset of the training data used in one pass of optimization.

  • Epoch: One complete cycle through the entire training dataset.

  • Iteration: A single update of model parameters.

Parameter

Model-specific values learned during training (e.g., weights in a neural network).

Hyperparameter

Values set before training that control the learning process (e.g., learning rate, number of layers).

Cost Function (Loss Function, Objective Function)

A function that measures how well a model's predictions match the actual outputs. Examples:

  • MSE for regression, Cross-Entropy Loss for classification.

Gradient Descent

An optimization algorithm is used to minimize the cost function by updating model parameters in the direction of the steepest descent.

Learning Rate

Controls the step size in updating model parameters during gradient descent.

Evaluation

The process of assessing a trained model’s performance using metrics like:

  • Accuracy, Precision, Recall, F1-Score (for classification).

  • RMSE, MAE (for regression).

Did you find this article valuable?

Support Nabaranjan's blog by becoming a sponsor. Any amount is appreciated!