Caching
A process of storing frequently accessed data to reduce latency and improve performance in machine learning applications.
Types of Caching
- Memory Caching - Stores data in RAM for quick access.
- Disk Caching - Uses SSDs or HDDs for larger but slower storage.
Example
Used in deep learning to cache dataset batches during training.
Capsule Networks
An advanced neural network architecture designed to better capture spatial hierarchies in images.
Types of Capsule Networks
- Dynamic Routing Capsules - Uses iterative routing.
- Matrix Capsules - Includes transformation matrices.
Example
Used in computer vision for improved object detection.
CatBoost
A gradient boosting algorithm optimized for categorical features.
Types of Boosting in CatBoost
- Ordered Boosting - Prevents target leakage.
- Plain Boosting - Faster but riskier for small datasets.
Example
Used in fraud detection models.
Causal Inference
A statistical approach to determine cause-and-effect relationships in data.
Types of Causal Inference
- Experimental Causal Inference - Uses randomized controlled trials.
- Observational Causal Inference - Uses statistical techniques like propensity scores.
Example
Used in economics to analyze policy impact.
Center Loss
A loss function that helps improve intra-class compactness in deep learning.
Types of Loss Functions Related to Center Loss
- Softmax Loss - Focuses on classification.
- Contrastive Loss - Maximizes inter-class distance.
Example
Used in face recognition systems.
Chain Rule in Machine Learning
A fundamental rule in calculus used to compute derivatives of composite functions.
Types of Chain Rule Applications
- Backpropagation - Uses chain rule to compute gradients.
- Probabilistic Chain Rule - Used in Bayesian networks.
Example
Used in training deep neural networks.
Character-Level Embeddings
A technique that represents words as sequences of character-level features instead of whole-word embeddings.
Types of Character-Level Embeddings
- Convolutional Character Embeddings - Uses CNNs to extract features.
- Recurrent Character Embeddings - Uses RNNs or LSTMs.
Example
Used in NLP tasks like spell correction.
Chi-Square Test
A statistical test used to determine if there is a significant relationship between categorical variables.
Types of Chi-Square Tests
- Chi-Square Goodness of Fit - Checks distribution similarity.
- Chi-Square Test of Independence - Examines relationships between variables.
Example
Used in feature selection for machine learning models.
Class Imbalance
A scenario where the distribution of target classes in a dataset is highly skewed.
Types of Class Imbalance Handling
- Oversampling - Increases minority class samples.
- Undersampling - Reduces majority class samples.
Example
Common in fraud detection datasets.
Clustering
An unsupervised learning technique that groups similar data points together.
Types of Clustering
- K-Means Clustering - Partitions data into K clusters.
- Hierarchical Clustering - Builds a tree of clusters.
Example
Used in customer segmentation for marketing.
Co-Training
A semi-supervised learning technique where two classifiers train each other using their most confident predictions.
Types of Co-Training
- Self-Training - Uses one classifier to label new data.
- Multi-View Learning - Uses multiple feature sets.
Example
Used in NLP for weakly supervised learning.
Collaborative Filtering
A recommendation system technique that predicts a user's interests based on past interactions.
Types of Collaborative Filtering
- Memory-Based - Uses similarity between users or items.
- Model-Based - Uses machine learning models like matrix factorization.
Example
Used in Netflix and Amazon recommendations.
Computational Learning Theory
A branch of machine learning that studies the theoretical properties of learning algorithms.
Types of Computational Learning
- PAC Learning - Probabilistically Approximately Correct learning.
- VC Dimension - Measures the capacity of a hypothesis class.
Example
Used in analyzing the efficiency of classifiers.
Concept Drift
A phenomenon where the statistical properties of a target variable change over time, affecting model performance.
Types of Concept Drift
- Sudden Drift - Abrupt changes in data patterns.
- Gradual Drift - Slow, continuous changes over time.
Example
Occurs in fraud detection models due to evolving attack patterns.
Types of Conformal Prediction
- Inductive Conformal Prediction - Uses a calibration set.
- Transductive Conformal Prediction - Uses test samples directly.
Example
Used in medical diagnostics to quantify uncertainty.
Conjugate Gradient Method
An optimization technique used for solving large-scale linear systems efficiently.
Types of Conjugate Gradient Methods
- Fletcher-Reeves - Standard method using gradients.
- Polak-Ribière - Variant with better convergence properties.
Example
Used in deep learning for optimizing weights.
Convex Optimization
A mathematical framework for minimizing convex functions, widely used in machine learning.
Types of Convex Optimization
- Linear Programming - Optimization with linear constraints.
- Quadratic Programming - Includes quadratic constraints.
Example
Used in support vector machines (SVMs).
Coreference Resolution
A natural language processing (NLP) task of determining when different words refer to the same entity.
Types of Coreference Resolution
- Pronominal Resolution - Resolving pronouns to their antecedents.
- Nominal Resolution - Linking named entities.
Example
Used in chatbot development.
Cross-Entropy Loss
A loss function used in classification tasks to measure the difference between predicted and true probability distributions.
Types of Cross-Entropy Loss
- Binary Cross-Entropy - Used for binary classification.
- Categorical Cross-Entropy - Used for multi-class classification.
Example
Used in deep learning models like CNNs and RNNs.
Curse of Dimensionality
A problem where increasing the number of features in a dataset makes it harder for models to learn effectively.
Types of Curse of Dimensionality
- Sparsity - Data points become more distant in high dimensions.
- Computational Complexity - Training models becomes slower.
Example
Occurs in high-dimensional text classification tasks.
CycleGAN
A type of Generative Adversarial Network (GAN) used for unpaired image-to-image translation.
Types of GANs Related to CycleGAN
- Pix2Pix - Requires paired training images.
- StarGAN - Can handle multiple domains.
Example
Used for converting satellite images to maps.
Cyclic Learning Rate
A learning rate policy that cyclically varies between a minimum and maximum value to accelerate training.
Types of Cyclic Learning Rate Schedules
- Triangular - Linear increase and decrease.
- Exponential - Exponential scaling up and down.
Example
Used in training deep neural networks.
Cyclic Redundancy Check (CRC)
An error-detection technique that verifies data integrity.
Types of CRC
- CRC-8 - Uses an 8-bit polynomial.
- CRC-32 - Uses a 32-bit polynomial for higher accuracy.
Example
Used in data transmission protocols.
CUDA (Compute Unified Device Architecture)
A parallel computing platform and API model developed by NVIDIA for GPU-accelerated computing.
Types of CUDA Optimizations
- Memory Coalescing - Improves memory access speed.
- Thread Parallelism - Enhances parallel execution.
Example
Used in deep learning frameworks like TensorFlow.
Cumulative Distribution Function (CDF)
A function that describes the probability that a random variable is less than or equal to a given value.
Types of CDF
- Empirical CDF - Based on observed data.
- Theoretical CDF - Derived from probability distributions.
Example
Used in probability estimation models.
Curvature Regularization
A technique that penalizes sharp changes in model parameters to enhance smoothness.
Types of Regularization Related to Curvature
- Laplacian Regularization - Penalizes second-order derivatives.
- Total Variation Regularization - Minimizes abrupt changes in model output.
Example
Used in image denoising models.
Contrastive Divergence
An optimization method used to train energy-based models like Restricted Boltzmann Machines (RBMs).
Types of Contrastive Divergence Variants
- CD-1 - Single-step contrastive divergence.
- Persistent Contrastive Divergence (PCD) - Uses a persistent Markov chain.
Example
Used in deep belief networks.
Conditional Random Fields (CRF)
A probabilistic model used for structured prediction tasks like sequence labeling.
Types of CRFs
- Linear Chain CRF - Used for sequence data.
- Skip Chain CRF - Allows long-range dependencies.
Example
Used in Named Entity Recognition (NER).
Confidence Interval
A statistical range that estimates an unknown parameter with a given probability.
Types of Confidence Intervals
- Parametric - Assumes data follows a known distribution.
- Non-Parametric - Makes fewer distributional assumptions.
Example
Used in A/B testing for decision-making.
Contrastive Loss
A loss function that helps learn similar representations for related data points and dissimilar ones for different data points.
Types of Contrastive Loss Functions
- Triplet Loss - Uses anchor, positive, and negative samples.
- InfoNCE - Used in contrastive self-supervised learning.
Example
Used in face recognition models.
Contrastive Representation Learning
A self-supervised learning approach that trains models by pulling similar data points closer and pushing dissimilar ones apart.
Types of Contrastive Learning
- Instance Discrimination - Compares individual examples.
- Cluster-Based - Groups similar examples together.
Example
Used in SimCLR and MoCo models for unsupervised learning.
Control Variates
A variance reduction technique in Monte Carlo simulations used to improve estimation accuracy.
Types of Control Variates
- Linear Control Variates - Uses known expectations.
- Non-Linear Control Variates - Uses complex dependencies.
Example
Used in reinforcement learning to stabilize training.
ConvLSTM (Convolutional LSTM)
A deep learning architecture combining convolutional and LSTM layers for spatiotemporal data.
Types of ConvLSTM Applications
- Video Prediction - Forecasts future frames.
- Weather Forecasting - Predicts atmospheric patterns.
Example
Used in autonomous driving for trajectory prediction.
Convolutional Autoencoder
An autoencoder variant that uses convolutional layers for unsupervised learning of image representations.
Types of Autoencoders
- Variational Autoencoder (VAE) - Learns probabilistic representations.
- Sparse Autoencoder - Enforces sparsity constraints.
Example
Used in image denoising tasks.
Convolutional Neural Network (CNN)
A deep learning model designed for image and spatial data processing using convolutional layers.
Types of CNN Architectures
- AlexNet - Early CNN with multiple convolutional layers.
- ResNet - Uses residual connections for deep networks.
Example
Used in facial recognition systems.
CoreML
Apple’s machine learning framework for deploying models on iOS devices.
Types of CoreML Models
- Neural Networks - Supports deep learning models.
- Tree Ensembles - Optimized for decision trees.
Example
Used in real-time object detection on iPhones.
Correlation Coefficient
A statistical measure of the relationship between two variables.
Types of Correlation Coefficients
- Pearson Correlation - Measures linear relationships.
- Spearman Correlation - Measures rank-based relationships.
Example
Used in feature selection for predictive models.
Cost Function
A mathematical function that measures the error between predicted and actual values in machine learning models.
Types of Cost Functions
- Mean Squared Error (MSE) - Used for regression.
- Cross-Entropy Loss - Used for classification.
Example
Used in gradient descent optimization.
Critical Learning Period
A concept from neuroscience referring to the time window when learning occurs most effectively.
Types of Critical Learning Period Applications
- Language Learning - Optimal period for language acquisition.
- Neural Network Training - Early-stage weight initialization.
Example
Used in transfer learning for knowledge retention.
Crossover in Genetic Algorithms
A genetic algorithm operation that combines two parent solutions to generate offspring.
Types of Crossover Techniques
- Single-Point Crossover - Swaps genetic material at one point.
- Uniform Crossover - Randomly selects genes from each parent.
Example
Used in evolutionary computing for optimization problems.
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.