Radial Basis Function (RBF)
A function used in machine learning and statistics to model complex relationships between inputs.
Types of RBF
- Gaussian RBF - Uses a Gaussian function for smooth interpolation.
- Multiquadric RBF - Uses polynomial-based radial functions.
Example
Used in RBF networks and SVM kernels for non-linear classification.
Random Forest
An ensemble learning method that constructs multiple decision trees to improve accuracy and reduce overfitting.
Types of Random Forest
- Classification Random Forest - Used for categorical predictions.
- Regression Random Forest - Used for numerical predictions.
Example
Used in fraud detection and recommendation systems.
Rank-Based Learning
A learning paradigm that focuses on ranking instances instead of direct classification or regression.
Types of Rank-Based Learning
- Pairwise Ranking - Compares pairs of instances to determine order.
- Listwise Ranking - Considers an entire list of instances at once.
Example
Used in search engines and recommendation systems.
Recurrent Neural Networks (RNN)
A type of neural network designed for sequence data by maintaining a memory of previous inputs.
Types of RNNs
- Vanilla RNN - Basic recurrent structure.
- LSTM (Long Short-Term Memory) - Overcomes vanishing gradient problems.
Example
Used in speech recognition and text generation.
Reinforcement Learning
A machine learning paradigm where agents learn optimal actions by interacting with an environment.
Types of Reinforcement Learning
- Model-Free RL - Does not require a model of the environment.
- Model-Based RL - Uses a model to predict environment dynamics.
Example
Used in game AI and robotics.
Regularization
A technique used to prevent overfitting by adding a penalty term to the loss function.
Types of Regularization
- L1 Regularization (Lasso) - Encourages sparsity.
- L2 Regularization (Ridge) - Penalizes large weights.
Example
Used in linear regression and neural networks.
Regression Analysis
A statistical method used to model relationships between dependent and independent variables.
Types of Regression
- Linear Regression - Models linear relationships.
- Polynomial Regression - Models non-linear relationships.
Example
Used in predictive analytics and economic forecasting.
Reinforcement Learning Policy
A strategy that an RL agent follows to decide on the next action based on the current state.
Types of RL Policies
- Deterministic Policy - Always takes the same action in a given state.
- Stochastic Policy - Chooses actions based on probability distributions.
Example
Used in self-driving cars and robotic automation.
ReLU (Rectified Linear Unit)
An activation function that outputs the input directly if positive and zero otherwise.
Types of ReLU
- Standard ReLU - Outputs max(0, x).
- Leaky ReLU - Allows small negative outputs to address dying neurons.
Example
Used in deep learning models like CNNs and RNNs.
Residual Networks (ResNet)
A type of deep neural network that uses residual connections to improve training.
Types of ResNet
- ResNet-18 - A shallow variant.
- ResNet-50 - A deeper architecture for complex tasks.
Example
Used in image classification and object detection.
Reproducible Machine Learning
The practice of ensuring ML experiments can be repeated with the same results.
Types of Reproducibility
- Exact Reproducibility - Identical outputs with identical inputs.
- Statistical Reproducibility - Similar outputs within a confidence interval.
Example
Used in scientific research and model validation.
Resampling Techniques
Methods for creating new training samples from existing data to improve model robustness.
Types of Resampling
- Bootstrap - Sampling with replacement.
- Cross-Validation - Splitting data for training and testing.
Example
Used in model validation and variance estimation.
Reward Function
A function in reinforcement learning that assigns a numerical reward to an agent’s action.
Types of Reward Functions
- Sparse Reward - Rewards given at specific milestones.
- Dense Reward - Rewards given continuously based on performance.
Example
Used in robotics and autonomous systems.
Ridge Regression
A type of regression that adds L2 regularization to reduce overfitting.
Types of Ridge Regression
- Standard Ridge Regression - Uses squared penalty term.
- Generalized Ridge Regression - Adapts weights based on prior knowledge.
Example
Used in finance and economic forecasting.
Robust Statistics
Statistical techniques that remain accurate despite outliers or noisy data.
Types of Robust Statistics
- Huber Regression - Handles outliers better than standard regression.
- Median Absolute Deviation (MAD) - Measures dispersion in skewed data.
Example
Used in medical data analysis and fraud detection.
ROC Curve (Receiver Operating Characteristic)
A graphical representation of a model’s classification performance across different thresholds.
Types of ROC Analysis
- Binary ROC Curve - For two-class classification.
- Multi-Class ROC - Extends ROC to multi-class problems.
Example
Used in medical diagnostics and credit scoring.
Root Mean Squared Error (RMSE)
A common metric to measure the average magnitude of model prediction errors.
Types of RMSE Applications
- Standard RMSE - Measures absolute prediction errors.
- Normalized RMSE - Scales RMSE for better comparability.
Example
Used in regression analysis and time-series forecasting.
Rule-Based Learning
A method of AI where models learn explicit IF-THEN rules from data.
Types of Rule-Based Learning
- Expert Systems - Uses predefined human knowledge.
- Inductive Rule Learning - Extracts rules automatically from data.
Example
Used in fraud detection and medical diagnosis.
Robust Machine Learning
ML models that maintain performance even with noisy or adversarial data.
Types of Robust ML
- Adversarial Robustness - Models resistant to adversarial attacks.
- Noise Robustness - Models handling incomplete or corrupt data.
Example
Used in cybersecurity and self-driving cars.
Reinforcement Learning Exploration-Exploitation Tradeoff
The balance between exploring new strategies and exploiting known rewards in RL.
Types of Tradeoff Strategies
- Epsilon-Greedy - Chooses random actions with probability ε.
- Upper Confidence Bound (UCB) - Prioritizes actions with high uncertainty.
Example
Used in game AI and robotic path planning.
Random Search
A hyperparameter tuning method that randomly selects values within a defined range.
Types of Hyperparameter Search
- Grid Search - Systematic, exhaustive search.
- Bayesian Optimization - Probabilistic model-based optimization.
Example
Used in deep learning model optimization.
Random Variable in ML
A variable whose possible values are outcomes of a probabilistic experiment.
Types of Random Variables
- Discrete Random Variable - Takes distinct values.
- Continuous Random Variable - Takes any value within an interval.
Example
Used in probabilistic models like Bayesian Networks.
Recurrent Neural Networks (RNN)
A type of neural network designed for sequential data.
Types of RNNs
- Vanilla RNN - Simple recurrent structure.
- LSTM (Long Short-Term Memory) - Handles long-term dependencies.
Example
Used in speech recognition and time-series forecasting.
Reinforcement Learning Policy
A strategy that an RL agent follows to decide actions.
Types of RL Policies
- Deterministic Policy - Always selects the best-known action.
- Stochastic Policy - Selects actions based on probability distributions.
Example
Used in robotics and automated trading.
Reinforcement Learning Value Function
A function estimating future rewards an agent will receive from a state.
Types of Value Functions
- State-Value Function (V) - Value of a given state.
- Action-Value Function (Q) - Value of a given action in a state.
Example
Used in Deep Q-Networks (DQN).
Reinforcement Learning Model-Free vs Model-Based
Two approaches in RL where model-free methods learn from interaction, and model-based methods use an environment model.
Types of RL Approaches
- Model-Free - Q-learning, Policy Gradient.
- Model-Based - AlphaGo, Dyna-Q.
Example
Used in AI for board games and autonomous driving.
Representation Learning
A learning method where features are automatically learned from raw data.
Types of Representation Learning
- Supervised Representation Learning - Features are guided by labeled data.
- Unsupervised Representation Learning - Features are extracted without labels.
Example
Used in deep learning for image recognition.
Residual Networks (ResNet)
A deep neural network architecture that introduces shortcut connections to prevent vanishing gradients.
Types of ResNet
- ResNet-18 - A smaller model suitable for edge computing.
- ResNet-50 - A deeper model used in large-scale image recognition.
Example
Used in ImageNet classification tasks.
Reinforcement Learning Temporal Difference Learning
A method of RL that updates value estimates based on new observations.
Types of TD Learning
- TD(0) - Updates values using immediate next-step rewards.
- TD(λ) - Generalizes TD(0) for multi-step updates.
Example
Used in game-playing AI like AlphaZero.
ReLU (Rectified Linear Unit)
A commonly used activation function in deep learning.
Types of ReLU Variants
- Leaky ReLU - Allows small negative values.
- PReLU (Parametric ReLU) - Learns the negative slope.
Example
Used in CNNs for feature extraction.
Restricted Boltzmann Machine (RBM)
A type of neural network used for unsupervised learning and feature extraction.
Types of RBM
- Binary RBM - Works with binary-valued inputs.
- Gaussian RBM - Works with continuous-valued inputs.
Example
Used in deep belief networks for dimensionality reduction.
Ridge Regression
A regression technique that adds a penalty term to reduce overfitting.
Types of Regularization
- Lasso Regression - Uses L1 penalty.
- Elastic Net - Combines L1 and L2 penalties.
Example
Used in financial modeling to predict stock prices.
Robust Scaling
A data preprocessing technique that scales data using medians and quartiles.
Types of Scaling
- Min-Max Scaling - Scales data between 0 and 1.
- Standard Scaling - Centers data around mean 0 and variance 1.
Example
Used when dealing with outliers in datasets.
ROC Curve (Receiver Operating Characteristic)
A graphical representation of a classifier’s performance at different thresholds.
Key Metrics
- True Positive Rate (TPR) - Sensitivity.
- False Positive Rate (FPR) - 1 - Specificity.
Example
Used in medical diagnostics for evaluating models.
Root Mean Square Error (RMSE)
A metric to measure the difference between predicted and actual values.
Comparison with Other Metrics
- Mean Absolute Error (MAE) - Measures absolute differences.
- Mean Squared Error (MSE) - Similar to RMSE but without square root.
Example
Used in evaluating regression models.
RProp (Resilient Backpropagation)
An optimization algorithm that adjusts weight updates independently.
Types of RProp
- iRProp+ - Improved with adaptive step sizes.
- RProp- - Standard version for backpropagation.
Example
Used in neural networks for fast training.
Rule-Based Machine Learning
A system where decisions are made based on predefined rules.
Types of Rule-Based Learning
- Decision Trees - Uses if-else rules.
- Expert Systems - Uses human-defined rules.
Example
Used in fraud detection systems.
Recursive Feature Elimination (RFE)
A feature selection technique that recursively removes less important features.
Types of Feature Selection
- Filter Methods - Uses statistical tests.
- Wrapper Methods - Uses model performance.
Example
Used in improving model accuracy.
Robust Regression
A regression method that reduces the effect of outliers.
Types of Robust Regression
- Huber Regression - Switches between MSE and MAE.
- RANSAC - Randomly samples data to fit models.
Example
Used in real-world noisy datasets.
Recurrent Convolutional Neural Network (RCNN)
A hybrid neural network combining CNNs and RNNs.
Types of RCNN
- RCNN with LSTM - Uses LSTM for sequence processing.
- RCNN with GRU - Uses GRU for faster training.
Example
Used in text and image recognition.
Machine Learning (ML)
ML is a subset of AI that enables machines to learn patterns from data and make predictions or decisions without explicit programming.
Types of ML
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
Example
Spam detection in emails using classification models.
Deep Learning (DL)
DL is a subset of ML that uses artificial neural networks to process complex data and perform high-level computations.
Example
Image recognition in self-driving cars.
Generative AI (Gen AI)
Gen AI refers to AI models that generate new content, including text, images, and code, using trained knowledge bases.
Example
AI models like ChatGPT and Stable Diffusion that generate text and images.