Glossary
**False Negative**: A type of error in a binary classification where the model incorrectly predicts the negative class for an instance that actually belongs to the positive class.
**False Positive**: A type of error in a binary classification where the model incorrectly predicts the positive class for an instance that actually belongs to the negative class.
**Feature Engineering**: The process of selecting, modifying, or creating new features from raw data to improve the performance of a machine learning model.
**Feature Extraction**: The process of transforming raw data into a set of features that can be used in modeling, often by reducing the data's dimensionality while retaining essential information.
**Feature Importance**: A technique used to quantify the contribution of each feature to the predictions made by a machine learning model, often used to improve interpretability and model performance.
**Feature Map**: The output of a convolutional layer in a Convolutional Neural Network (CNN), representing the activation of different filters applied to the input data.
**Feature Scaling**: The process of normalizing or standardizing features in a dataset so that they contribute equally to the model, often necessary for algorithms that rely on distance metrics.
**Federated Learning**: A decentralized approach to machine learning where models are trained on data distributed across multiple devices or servers without sharing the data itself.
**Feedforward Neural Network**: A type of neural network where connections between nodes do not form a cycle, typically used in simple classification and regression tasks.
**Fidelity**: The accuracy or precision of a model's predictions relative to the true values or outcomes, often used to assess the quality of a machine learning model.
**Fisher’s Exact Test**: A statistical significance test used to determine if there are nonrandom associations between two categorical variables, often used in small sample sizes.
**Few-Shot Learning**: A machine learning approach where models are trained to generalize well from only a few examples, often used in tasks where labeled data is scarce.
**F1 Score**: A metric that combines precision and recall into a single value, calculated as the harmonic mean of precision and recall, used to evaluate the performance of a classification model.
**Forward Propagation**: The process in a neural network where inputs are passed through the network's layers to generate an output, typically followed by backpropagation for training.
**Fuzzy Logic**: A form of logic where truth values are expressed in degrees rather than as absolute true or false, often used in systems that require reasoning with uncertainty.
**Fuzzy Clustering**: A clustering technique where data points can belong to more than one cluster with varying degrees of membership, allowing for more flexible grouping.
**Full-Stack Data Scientist**: A data scientist with expertise across the entire data science pipeline, including data collection, preprocessing, modeling, deployment, and monitoring.
**Functional Programming**: A programming paradigm that treats computation as the evaluation of mathematical functions and avoids changing state or mutable data, often used in developing robust and testable machine learning code.
**Frequentist Inference**: A statistical approach that interprets probability as the frequency of events occurring over repeated experiments, as opposed to Bayesian inference, which incorporates prior beliefs.
**Feature Selection**: The process of identifying the most relevant features in a dataset to use in building a machine learning model, often used to improve model performance and reduce overfitting.
**Fourier Transform**: A mathematical transformation that decomposes a function or signal into its constituent frequencies, often used in signal processing and time series analysis.
**Flat Clustering**: A type of clustering algorithm that partitions data into a set of distinct clusters without any hierarchical structure, commonly used in k-means clustering.
**Facial Recognition**: A technology that identifies or verifies a person by analyzing facial features from an image or video, often used in security and identification systems.
**Forward Selection**: A feature selection technique that starts with an empty model and adds features one by one based on their contribution to the model's performance.
**Fixed-Point Iteration**: A method of computing fixed points of a function by iteratively applying the function to an initial guess, often used in numerical analysis and optimization.
**Formal Verification**: The process of proving or disproving the correctness of algorithms with respect to a certain formal specification using formal methods of mathematics.
**Factor Analysis**: A statistical method used to identify underlying relationships between variables by modeling the data in terms of a smaller number of unobserved factors.
**Fractal Dimension**: A measure of the complexity of a shape or pattern, often used in image analysis and computer vision to quantify texture and irregularity.
**Feature Space**: The multidimensional space defined by the features used in a machine learning model, where each dimension corresponds to a different feature.
**Forward-Backward Algorithm**: An algorithm used in hidden Markov models to compute the probabilities of a sequence of observations given a model, often used in speech recognition and bioinformatics.
**Focal Loss**: A loss function that down-weights well-classified examples, enabling the model to focus on hard-to-classify cases, often used in object detection tasks.
**Fibonacci Search**: A search algorithm that reduces the search space by a factor related to the Fibonacci sequence, often used in optimization problems.
**Flow-based Generative Models**: A class of generative models that use invertible neural networks to map data to a latent space, enabling efficient sampling and likelihood estimation.
**Factorization Machine**: A machine learning model that generalizes matrix factorization techniques to higher-order interactions, often used in recommendation systems.
**Fixed-Point Arithmetic**: A method of representing fractional numbers using integer arithmetic, often used in hardware implementations of machine learning models where floating-point operations are expensive.
**Free Energy Principle**: A theoretical framework in neuroscience and machine learning that describes how systems minimize uncertainty by updating their internal models based on new information.
**Fine-Tuning**: The process of adapting a pre-trained model to a new task by training it on a smaller, task-specific dataset, often used in transfer learning.
**Filter Method**: A feature selection technique that selects features based on their statistical properties, often used as a preprocessing step before model training.
**Fairness in AI**: The practice of designing and deploying AI systems in a way that avoids bias and ensures equitable outcomes across different groups of people.
**False Discovery Rate (FDR)**: The expected proportion of false positives among all significant results, often used in multiple hypothesis testing to control the likelihood of false positives.
**Fourier Series**: A way to represent a function as a sum of sinusoidal functions, often used in signal processing to analyze periodic functions.
**Functional API (Keras)**: A way to define models in Keras that allows for more flexible architectures, such as models with shared layers or multiple inputs and outputs.
**Frobenius Norm**: A matrix norm that is the square root of the sum of the absolute squares of its elements, often used in optimization and regularization.
**Factor Graph**: A bipartite graph representing the factorization of a probability distribution, often used in graphical models and belief propagation algorithms.
**Friedman Test**: A non-parametric statistical test used to detect differences in treatments across multiple test attempts, often used in machine learning model evaluation.
**Feature Pyramid Network (FPN)**: A type of neural network architecture that builds feature pyramids for detecting objects at different scales, often used in object detection tasks.
**Function Approximation**: The process of estimating a function that best fits a set of data points, often used in machine learning to generalize from training data to unseen examples.
**Fisher Vector**: A representation that combines the generative model of a dataset with a discriminative classifier, often used in image classification tasks.