# Edge AI & Industrial IoT Glossary ## metadata - title: Edge AI Glossary | Klyff - route: /resources/glossary - total_terms: 111 - format: directory_index ## Terminology Catalog ### [Accelerators](/resources/glossary/accelerators) - **Letter:** A - **Definition:** Accelerators are specialized hardware components that speed up AI model computations at the Edge. These processing units (such as GPUs, TPUs, and NPUs) enhance performance by... links: [/agent/glossary/accelerators.md] ### [Accuracy](/resources/glossary/accuracy) - **Letter:** A - **Definition:** Accuracy in Edge AI refers to how closely an AI model’s predictions match real-world outcomes. It measures the precision and reliability of edge-deployed algorithms in processing... links: [/agent/glossary/accuracy.md] ### [Active Learning](/resources/glossary/active-learning) - **Letter:** A - **Definition:** Active Learning in Edge AI is a machine learning technique where models selectively query the most informative data for labeling. This human-in-the-loop approach helps AI systems... links: [/agent/glossary/active-learning.md] ### [Algorithm Bias](/resources/glossary/algorithm-bias) - **Letter:** A - **Definition:** Algorithm bias in Edge AI refers to systematic errors in decision-making models caused by skewed training data or flawed assumptions. In simple terms, it happens when an AI system... links: [/agent/glossary/algorithm-bias.md] ### [Anomaly Detection](/resources/glossary/anomaly-detection) - **Letter:** A - **Definition:** Anomaly Detection is the process of identifying patterns or data points that deviate from normal behavior in real time. In the context of Edge AI, it enables intelligent devices... links: [/agent/glossary/anomaly-detection.md] ### [Application-Specific Integrated Circuits](/resources/glossary/application-specific-integrated-circuits) - **Letter:** A - **Definition:** Application-Specific Integrated Circuits (ASICs) are custom-designed semiconductor chips built to perform a specific task or function efficiently. In Edge AI, ASICs deliver... links: [/agent/glossary/application-specific-integrated-circuits.md] ### [Artificial Intelligence](/resources/glossary/artificial-intelligence) - **Letter:** A - **Definition:** Artificial Intelligence (AI) is the capability of machines to perform tasks that typically require human intelligence, such as learning, reasoning, and problem-solving. In Edge AI,... links: [/agent/glossary/artificial-intelligence.md] ### [Automated Machine Learning](/resources/glossary/automated-machine-learning) - **Letter:** A - **Definition:** Automated Machine Learning (AutoML) automates the process of building, training, and optimizing machine learning models, making AI accessible even without deep data science... links: [/agent/glossary/automated-machine-learning.md] ### [Bias](/resources/glossary/bias) - **Letter:** B - **Definition:** Bias is a systematic error or preference that skews data, algorithms, or human decisions away from neutrality. In AI and machine learning, bias - sometimes called prejudice or... links: [/agent/glossary/bias.md] ### [Black Box](/resources/glossary/black-box) - **Letter:** B - **Definition:** A Black Box in Edge AI refers to an AI model or system whose internal decision-making process is not transparent or easily understood. It takes inputs and produces outputs without... links: [/agent/glossary/black-box.md] ### [BLERP](/resources/glossary/blerp) - **Letter:** B - **Definition:** BLERP is an advanced Edge AI framework that enables real-time data analysis and decision-making directly on edge devices. Also known as Edge AI inference, BLERP reduces latency,... links: [/agent/glossary/blerp.md] ### [Brownfield Project](/resources/glossary/brownfield-project) - **Letter:** B - **Definition:** A Brownfield Project in Edge AI refers to upgrading or integrating AI and IoT solutions into existing industrial or enterprise infrastructure. Also called a retrofit deployment,... links: [/agent/glossary/brownfield-project.md] ### [Cascading Flow](/resources/glossary/cascading-flow) - **Letter:** C - **Definition:** Cascade, or Cascading Flow in Edge AI refers to the sequential processing of data across multiple edge devices or layers, enabling faster, efficient, and context-aware... links: [/agent/glossary/cascading-flow.md] ### [Classification](/resources/glossary/classification) - **Letter:** C - **Definition:** Classification in Edge AI is the process of categorizing data into predefined groups or labels using on-device machine learning models. Also known as data labeling or... links: [/agent/glossary/classification.md] ### [Cloud Computing](/resources/glossary/cloud-computing) - **Letter:** C - **Definition:** Cloud computing is the delivery of computing resources (such as storage, processing, and networking) over the internet, enabling devices and applications to operate without local... links: [/agent/glossary/cloud-computing.md] ### [Clustering](/resources/glossary/clustering) - **Letter:** C - **Definition:** Clustering is a technique in Edge AI that organizes data or devices into meaningful groups based on patterns or similarities. It allows local devices to analyze data efficiently,... links: [/agent/glossary/clustering.md] ### [Computer Vision](/resources/glossary/computer-vision) - **Letter:** C - **Definition:** Computer vision, a core Edge AI technology, enables devices to interpret and analyze visual data from cameras or sensors in real time. Also known as machine vision, it allows edge... links: [/agent/glossary/computer-vision.md] ### [Conditionals and Heuristics](/resources/glossary/conditionals-and-heuristics) - **Letter:** C - **Definition:** Conditionals and heuristics in Edge AI are rules and shortcuts used by edge devices to make fast, efficient decisions without cloud dependency. By applying logical conditions or... links: [/agent/glossary/conditionals-and-heuristics.md] ### [Confusion Matrix](/resources/glossary/confusion-matrix) - **Letter:** C - **Definition:** A confusion matrix is a tabular tool used in Edge AI to evaluate the accuracy of machine learning models by comparing predicted outcomes against actual results. Also called an... links: [/agent/glossary/confusion-matrix.md] ### [Containerization](/resources/glossary/containerization) - **Letter:** C - **Definition:** Containerization is a method of bundling AI applications with all required libraries and configurations into isolated containers. This ensures seamless execution on edge devices... links: [/agent/glossary/containerization.md] ### [Convolutional Models](/resources/glossary/convolutional-models) - **Letter:** C - **Definition:** Convolutional models, also known as convolutional neural networks (CNNs), are specialized AI algorithms designed to process and analyze visual data at the edge. These models... links: [/agent/glossary/convolutional-models.md] ### [Cross-Validation](/resources/glossary/cross-validation) - **Letter:** C - **Definition:** Cross-validation is a statistical method used to assess how well an AI or machine learning model will generalize to unseen data. In Edge AI, it helps ensure models deployed on... links: [/agent/glossary/cross-validation.md] ### [Data Augmentation](/resources/glossary/data-augmentation) - **Letter:** D - **Definition:** Data Augmentation is a technique in Edge AI that artificially increases the volume and diversity of datasets by transforming existing data. Also called data synthesis, it improves... links: [/agent/glossary/data-augmentation.md] ### [Data Latency](/resources/glossary/data-latency) - **Letter:** D - **Definition:** Data latency, also called information delay, is the time gap between data generation at the edge and its availability for processing. In Edge AI, minimizing data latency ensures... links: [/agent/glossary/data-latency.md] ### [Data Pipeline](/resources/glossary/data-pipeline) - **Letter:** D - **Definition:** A data pipeline is a structured system that moves and processes data from edge devices to storage or AI models. Also called a data workflow, it enables real-time data collection,... links: [/agent/glossary/data-pipeline.md] ### [Data-Centric Machine Learning](/resources/glossary/data-centric-machine-learning) - **Letter:** D - **Definition:** Data-Centric Machine Learning (DCML) focuses on improving AI models by refining and curating the data rather than modifying the algorithms. In Edge AI, this approach ensures... links: [/agent/glossary/data-centric-machine-learning.md] ### [Deep Learning](/resources/glossary/deep-learning) - **Letter:** D - **Definition:** Deep learning, a subset of artificial intelligence (AI), enables Edge AI devices to learn from data patterns and make intelligent decisions locally. Also called neural network... links: [/agent/glossary/deep-learning.md] ### [Digital Signal Processing](/resources/glossary/digital-signal-processing) - **Letter:** D - **Definition:** Digital Signal Processing (DSP) is the analysis and manipulation of digital signals to improve, interpret, or optimize data. In Edge AI, DSP enables real-time processing of audio,... links: [/agent/glossary/digital-signal-processing.md] ### [Drift](/resources/glossary/drift) - **Letter:** D - **Definition:** Drift in Edge AI refers to the gradual change in data patterns or AI model performance when deployed on edge devices. Also called model drift, it impacts decision accuracy and... links: [/agent/glossary/drift.md] ### [Duty Cycle](/resources/glossary/duty-cycle) - **Letter:** D - **Definition:** A duty cycle in Edge AI refers to the percentage of time an edge device or sensor is actively performing tasks versus idle time. Also called operational cycle, it helps optimize... links: [/agent/glossary/duty-cycle.md] ### [Edge](/resources/glossary/edge) - **Letter:** E - **Definition:** Edge, or edge computing, is the practice of processing data locally on devices rather than sending it to centralized cloud servers. In Edge AI, this allows AI algorithms to... links: [/agent/glossary/edge.md] ### [Edge AI](/resources/glossary/edge-ai) - **Letter:** E - **Definition:** Edge AI is the deployment of artificial intelligence (AI) directly on edge devices (like sensors, cameras, or IoT hardware) allowing real-time data processing without relying on the... links: [/agent/glossary/edge-ai.md] ### [Edge Computing](/resources/glossary/edge-computing) - **Letter:** E - **Definition:** Edge computing, also called distributed computing, is the practice of processing data closer to the source (such as IoT devices) rather than relying solely on centralized cloud... links: [/agent/glossary/edge-computing.md] ### [Edge Devices](/resources/glossary/edge-devices) - **Letter:** E - **Definition:** Edge devices are hardware units, such as sensors, gateways, or IoT appliances, that process data locally rather than sending it to a centralized cloud. Also called edge nodes,... links: [/agent/glossary/edge-devices.md] ### [Embedded Machine Learning](/resources/glossary/embedded-machine-learning) - **Letter:** E - **Definition:** Embedded Machine Learning (EML) brings AI models directly onto edge devices, enabling real-time data processing without relying on cloud connectivity. Also known as on-device AI,... links: [/agent/glossary/embedded-machine-learning.md] ### [Embedded Systems](/resources/glossary/embedded-systems) - **Letter:** E - **Definition:** Embedded systems are specialized computing systems designed to perform dedicated tasks within larger devices. In the context of Edge AI, these systems enable real-time processing... links: [/agent/glossary/embedded-systems.md] ### [Emulator](/resources/glossary/emulator) - **Letter:** E - **Definition:** An emulator is a software tool that mimics the behavior of an Edge AI device or system, allowing developers to test and optimize AI models without needing the physical hardware.... links: [/agent/glossary/emulator.md] ### [Ensemble](/resources/glossary/ensemble) - **Letter:** E - **Definition:** In Edge AI, an ensemble is a group of multiple AI models working together to improve prediction accuracy and decision-making. Also called model aggregation, it leverages diverse... links: [/agent/glossary/ensemble.md] ### [Explainability](/resources/glossary/explainability) - **Letter:** E - **Definition:** Explainability in Edge AI refers to the ability to interpret and understand how AI models running on edge devices make decisions. Also called AI interpretability, it ensures... links: [/agent/glossary/explainability.md] ### [F1 Score](/resources/glossary/f1-score) - **Letter:** F - **Definition:** The F1 Score, also called the F-measure, is a key performance metric in Edge AI that balances precision and recall to evaluate model accuracy, especially for tasks with uneven... links: [/agent/glossary/f1-score.md] ### [Fairness](/resources/glossary/fairness) - **Letter:** F - **Definition:** Fairness in Edge AI ensures that algorithms and models make unbiased, equitable decisions, minimizing discrimination or favoritism. It promotes ethical, reliable AI outcomes and... links: [/agent/glossary/fairness.md] ### [Feature Engineering](/resources/glossary/feature-engineering) - **Letter:** F - **Definition:** Feature engineering is the process of transforming raw sensor or device data into meaningful input features that improve the performance of Edge AI models. It involves selecting,... links: [/agent/glossary/feature-engineering.md] ### [Federated Learning](/resources/glossary/federated-learning) - **Letter:** F - **Definition:** Federated Learning is a decentralized machine learning approach that trains AI models across multiple edge devices without sharing raw data. Instead of centralizing data, it sends... links: [/agent/glossary/federated-learning.md] ### [Field Programmable Gate Array](/resources/glossary/field-programmable-gate-array) - **Letter:** F - **Definition:** A Field Programmable Gate Array (FPGA) is a reconfigurable semiconductor device that allows developers to customize hardware functions after manufacturing. In Edge AI, FPGAs... links: [/agent/glossary/field-programmable-gate-array.md] ### [Firmware](/resources/glossary/firmware) - **Letter:** F - **Definition:** Firmware is a specialized form of software embedded directly into hardware devices, enabling them to function and communicate intelligently. In Edge AI systems, firmware bridges... links: [/agent/glossary/firmware.md] ### [Flash Memory](/resources/glossary/flash-memory) - **Letter:** F - **Definition:** Flash memory is a non-volatile storage technology that retains data without power, making it ideal for Edge AI devices and IoT systems. It stores critical data, models, and... links: [/agent/glossary/flash-memory.md] ### [Framing](/resources/glossary/framing) - **Letter:** F - **Definition:** Framing in Edge AI refers to the process of structuring raw sensor or device data into meaningful frames that AI models can analyze in real time. This data segmentation, also... links: [/agent/glossary/framing.md] ### [Generalization](/resources/glossary/generalization) - **Letter:** G - **Definition:** In Edge AI, generalization refers to an AI model’s ability to perform accurately on unseen data, not just the examples it was trained on. It ensures that edge devices can make... links: [/agent/glossary/generalization.md] ### [Graphics Processing Unit](/resources/glossary/graphics-processing-unit) - **Letter:** G - **Definition:** A Graphics Processing Unit (GPU) is a specialized processor designed to handle complex mathematical computations, primarily for rendering graphics and accelerating AI workloads.... links: [/agent/glossary/graphics-processing-unit.md] ### [Greenfield Project](/resources/glossary/greenfield-project) - **Letter:** G - **Definition:** A Greenfield Project in Edge AI refers to developing a completely new system, infrastructure, or deployment without constraints from existing architecture. Unlike brownfield... links: [/agent/glossary/greenfield-project.md] ### [Heterogeneous Compute](/resources/glossary/heterogeneous-compute) - **Letter:** H - **Definition:** Heterogeneous compute, also known as hybrid computing, refers to the use of different types of processors (like CPUs, GPUs, and AI accelerators) working together to handle diverse... links: [/agent/glossary/heterogeneous-compute.md] ### [Heuristics](/resources/glossary/heuristics) - **Letter:** H - **Definition:** Heuristics in Edge AI are simplified decision-making methods or “rules of thumb” that enable AI models to make quick, efficient choices at the edge without needing full data... links: [/agent/glossary/heuristics.md] ### [Human-in-the-Loop](/resources/glossary/human-in-the-loop) - **Letter:** H - **Definition:** Human-in-the-Loop (HITL) is a process in Edge AI where human feedback or intervention is integrated into machine learning systems to improve decision-making accuracy. Also known... links: [/agent/glossary/human-in-the-loop.md] ### [Inference](/resources/glossary/inference) - **Letter:** I - **Definition:** Inference in Edge AI refers to the process where a trained machine learning model makes real-time predictions or decisions directly on a local device without relying on cloud... links: [/agent/glossary/inference.md] ### [Internet of Things](/resources/glossary/internet-of-things) - **Letter:** I - **Definition:** The Internet of Things (IoT) refers to a network of connected devices (sensors, machines, and systems) that collect and exchange data in real time. When combined with Edge AI, IoT... links: [/agent/glossary/internet-of-things.md] ### [Interpreter](/resources/glossary/interpreter) - **Letter:** I - **Definition:** An interpreter or runtime in Edge AI is a lightweight software layer that executes machine learning models directly on edge devices without precompilation. It translates... links: [/agent/glossary/interpreter.md] ### [Kernel](/resources/glossary/kernel) - **Letter:** K - **Definition:** A kernel is the core component of an operating system or software framework that manages hardware communication, memory, and processes. In Edge AI, the kernel (also called the... links: [/agent/glossary/kernel.md] ### [Knowledge Distillation](/resources/glossary/knowledge-distillation) - **Letter:** K - **Definition:** Knowledge Distillation is a machine learning technique where a large, complex AI model (teacher) transfers its knowledge to a smaller, faster model (student) optimized for edge... links: [/agent/glossary/knowledge-distillation.md] ### [Label Noise](/resources/glossary/label-noise) - **Letter:** L - **Definition:** Label noise, also known as mislabeling or annotation errors, refers to inaccuracies in labeled datasets used to train Edge AI models. These errors can arise from human mistakes,... links: [/agent/glossary/label-noise.md] ### [Labels](/resources/glossary/labels) - **Letter:** L - **Definition:** Labels are descriptive tags or annotations assigned to data, enabling Edge AI models to understand, categorize, and make decisions from raw inputs. Also known as data tags, they... links: [/agent/glossary/labels.md] ### [Latency](/resources/glossary/latency) - **Letter:** L - **Definition:** Latency, also known as response delay, refers to the time it takes for data to travel from a device to a server and back. In Edge AI, low latency ensures real-time decision-making... links: [/agent/glossary/latency.md] ### [Lightweight Models](/resources/glossary/lightweight-models) - **Letter:** L - **Definition:** Lightweight models, also called compact AI models, are streamlined machine learning algorithms optimized to run efficiently on edge devices with limited computing power. They... links: [/agent/glossary/lightweight-models.md] ### [Local Processing](/resources/glossary/local-processing) - **Letter:** L - **Definition:** Local Processing, also known as on-device computation, is the practice of processing data directly on edge devices rather than sending it to a centralized server. This enables... links: [/agent/glossary/local-processing.md] ### [Loss](/resources/glossary/loss) - **Letter:** L - **Definition:** In Edge AI, loss is a metric that measures the difference between predicted outputs and actual results in machine learning models. Also called an error function, it helps quantify... links: [/agent/glossary/loss.md] ### [Machine Learning](/resources/glossary/machine-learning) - **Letter:** M - **Definition:** Machine Learning (ML) is a subset of artificial intelligence that enables devices to learn from data, identify patterns, and make decisions with minimal human intervention. In... links: [/agent/glossary/machine-learning.md] ### [Machine Learning Operations](/resources/glossary/machine-learning-operations) - **Letter:** M - **Definition:** Machine Learning Operations (MLOps) is the practice of deploying, monitoring, and managing machine learning models efficiently in production, particularly on Edge AI devices. Also... links: [/agent/glossary/machine-learning-operations.md] ### [Mean Average Precision](/resources/glossary/mean-average-precision) - **Letter:** M - **Definition:** Mean Average Precision (mAP) is a metric used to evaluate the accuracy of object detection models in Edge AI systems. Also known as mean precision, it quantifies how well an AI... links: [/agent/glossary/mean-average-precision.md] ### [Metadata](/resources/glossary/metadata) - **Letter:** M - **Definition:** Metadata is structured information that describes, organizes, and contextualizes data collected by Edge AI devices. Also called “data descriptors,” metadata makes raw sensor data,... links: [/agent/glossary/metadata.md] ### [Microcontroller Units](/resources/glossary/microcontroller-units) - **Letter:** M - **Definition:** Microcontroller Units (MCUs) are small, integrated computing devices that manage specific tasks in electronic systems. In Edge AI, they act as the “brain” of devices, processing... links: [/agent/glossary/microcontroller-units.md] ### [Model](/resources/glossary/model) - **Letter:** M - **Definition:** A model in Edge AI is a trained algorithm or digital blueprint that enables devices to make intelligent decisions locally, without cloud dependence. Also called an AI model, it... links: [/agent/glossary/model.md] ### [Model Cards](/resources/glossary/model-cards) - **Letter:** M - **Definition:** Model Cards are concise, standardized documents that describe an AI model’s key characteristics, performance metrics, and intended use cases. Also known as AI model datasheets,... links: [/agent/glossary/model-cards.md] ### [Neural Network](/resources/glossary/neural-network) - **Letter:** N - **Definition:** A neural network is a computational model inspired by the human brain that processes information through interconnected nodes, or “neurons.” In Edge AI, neural networks enable... links: [/agent/glossary/neural-network.md] ### [Neural Processing Unit](/resources/glossary/neural-processing-unit) - **Letter:** N - **Definition:** A Neural Processing Unit (NPU) is a specialized microprocessor designed to accelerate AI and deep learning computations, particularly neural network inference. Unlike CPUs or... links: [/agent/glossary/neural-processing-unit.md] ### [Object Detection](/resources/glossary/object-detection) - **Letter:** O - **Definition:** Object detection is an AI technique that identifies and locates multiple objects within an image or video in real time. In the context of Edge AI, it enables devices like cameras,... links: [/agent/glossary/object-detection.md] ### [On-Device Training](/resources/glossary/on-device-training) - **Letter:** O - **Definition:** On-device training is the process where AI models are trained and updated directly on edge devices (such as sensors, cameras, or IoT nodes) rather than in centralized cloud... links: [/agent/glossary/on-device-training.md] ### [Overfitting](/resources/glossary/overfitting) - **Letter:** O - **Definition:** Overfitting in Edge AI occurs when a machine learning model learns training data too precisely (including noise and irrelevant details) causing poor performance on new, unseen... links: [/agent/glossary/overfitting.md] ### [Parameters](/resources/glossary/parameters) - **Letter:** P - **Definition:** In Edge AI, parameters are the internal variables or weights within an AI model that determine how it processes and interprets data. Each parameter fine-tunes the model’s... links: [/agent/glossary/parameters.md] ### [Pose Estimation](/resources/glossary/pose-estimation) - **Letter:** P - **Definition:** Pose Estimation is an AI technique that identifies and tracks the position and orientation of objects or human body parts in real time. In Edge AI, it enables devices to interpret... links: [/agent/glossary/pose-estimation.md] ### [Postprocessing](/resources/glossary/postprocessing) - **Letter:** P - **Definition:** Postprocessing in Edge AI refers to the stage where raw inference results from an AI model are refined, filtered, or formatted into usable insights. It ensures that data... links: [/agent/glossary/postprocessing.md] ### [Precision](/resources/glossary/precision) - **Letter:** P - **Definition:** Precision in Edge AI refers to how accurately a model identifies true positive outcomes without false alarms. It measures the correctness of AI predictions made directly on edge... links: [/agent/glossary/precision.md] ### [Predictive Maintenance](/resources/glossary/predictive-maintenance) - **Letter:** P - **Definition:** Predictive Maintenance is a data-driven approach that uses Edge AI and machine learning algorithms to monitor equipment in real time, detect anomalies, and predict potential... links: [/agent/glossary/predictive-maintenance.md] ### [Pruning](/resources/glossary/pruning) - **Letter:** P - **Definition:** Pruning is a technique in Edge AI that removes unnecessary connections or parameters from machine learning models to improve efficiency without sacrificing accuracy. Also known as... links: [/agent/glossary/pruning.md] ### [Psychological Safety](/resources/glossary/psychological-safety) - **Letter:** P - **Definition:** Psychological safety is the belief that team members can share ideas, concerns, or mistakes without fear of negative consequences. In the context of Edge AI, it enables engineers... links: [/agent/glossary/psychological-safety.md] ### [Quantization](/resources/glossary/quantization) - **Letter:** Q - **Definition:** Quantization is the process of reducing the precision of a neural network’s weights and activations, converting them from high-bit formats (e.g., 32-bit floating point) to... links: [/agent/glossary/quantization.md] ### [Real-time Processing](/resources/glossary/real-time-processing) - **Letter:** R - **Definition:** Real-time processing is the immediate collection, analysis, and action on data as it’s generated, without delay. In Edge AI, it enables devices to make autonomous decisions... links: [/agent/glossary/real-time-processing.md] ### [Recall](/resources/glossary/recall) - **Letter:** R - **Definition:** Recall in Edge AI refers to the model’s ability to correctly identify all relevant positive instances from data processed at the edge. In simpler terms, it measures how many... links: [/agent/glossary/recall.md] ### [Receiver Operating Characteristic Curve](/resources/glossary/receiver-operating-characteristic-curve) - **Letter:** R - **Definition:** A Receiver Operating Characteristic (ROC) Curve is a graphical representation that evaluates the performance of an AI model by plotting the true positive rate (sensitivity)... links: [/agent/glossary/receiver-operating-characteristic-curve.md] ### [Regression](/resources/glossary/regression) - **Letter:** R - **Definition:** Regression is a core machine learning technique used in Edge AI to predict continuous values (like temperature, energy usage, or performance metrics) based on sensor or device data.... links: [/agent/glossary/regression.md] ### [Resource Constraints](/resources/glossary/resource-constraints) - **Letter:** R - **Definition:** Resource constraints in Edge AI refer to the limited availability of computing resources (such as processing power, memory, storage, and energy) on edge devices. Unlike cloud... links: [/agent/glossary/resource-constraints.md] ### [Responsible Design](/resources/glossary/responsible-design) - **Letter:** R - **Definition:** Responsible Design in Edge AI refers to the practice of developing AI-driven systems and devices that prioritize ethics, privacy, transparency, and sustainability. It ensures that... links: [/agent/glossary/responsible-design.md] ### [Sample Rate](/resources/glossary/sample-rate) - **Letter:** S - **Definition:** Sample rate refers to the number of data samples captured per second by a sensor or device. In Edge AI, it determines how frequently signals, such as audio, motion, or environmental... links: [/agent/glossary/sample-rate.md] ### [Segmentation](/resources/glossary/segmentation) - **Letter:** S - **Definition:** Segmentation in Edge AI refers to the process of dividing visual or sensory data into meaningful regions or objects for real-time analysis directly on edge devices. It helps... links: [/agent/glossary/segmentation.md] ### [Semi-supervised Learning](/resources/glossary/semi-supervised-learning) - **Letter:** S - **Definition:** Semi-supervised learning is an AI technique that combines a small amount of labeled data with a large pool of unlabeled data to train models more efficiently. In Edge AI, it helps... links: [/agent/glossary/semi-supervised-learning.md] ### [Sensor Fusion](/resources/glossary/sensor-fusion) - **Letter:** S - **Definition:** Sensor fusion, also called multi-sensor integration, is the process of merging data from multiple sensors to produce more accurate, reliable, and actionable insights. In Edge AI... links: [/agent/glossary/sensor-fusion.md] ### [Sequence Models](/resources/glossary/sequence-models) - **Letter:** S - **Definition:** Sequence models, also called sequential neural networks, are AI algorithms designed to analyze and predict data that unfolds over time. In Edge AI, they process real-time sensor,... links: [/agent/glossary/sequence-models.md] ### [Spectrogram](/resources/glossary/spectrogram) - **Letter:** S - **Definition:** A spectrogram is a visual representation of the frequency spectrum of signals over time, also called a frequency-time map. In Edge AI, spectrograms allow devices to analyze audio,... links: [/agent/glossary/spectrogram.md] ### [Spiking Neural Networks](/resources/glossary/spiking-neural-networks) - **Letter:** S - **Definition:** Spiking Neural Networks (SNNs) are a type of artificial neural network that mimic the way biological neurons communicate through discrete spikes. In Edge AI, they enable... links: [/agent/glossary/spiking-neural-networks.md] ### [Stakeholders](/resources/glossary/stakeholders) - **Letter:** S - **Definition:** In Edge AI, stakeholders are individuals, teams, or organizations who influence or are impacted by decisions, data insights, and AI-driven operations at the network edge. They are... links: [/agent/glossary/stakeholders.md] ### [Supervised Learning](/resources/glossary/supervised-learning) - **Letter:** S - **Definition:** Supervised learning is a type of machine learning where Edge AI systems are trained on labeled data to predict outcomes or classify information accurately. Also known as guided... links: [/agent/glossary/supervised-learning.md] ### [Synthetic Data](/resources/glossary/synthetic-data) - **Letter:** S - **Definition:** Synthetic data is artificially generated information that mimics real-world datasets, enabling AI models to learn without relying solely on sensitive or limited real data. Also... links: [/agent/glossary/synthetic-data.md] ### [System-on-Chip](/resources/glossary/system-on-chip) - **Letter:** S - **Definition:** A System-on-Chip (SoC) is an integrated circuit that combines multiple components (CPU, GPU, memory, and AI accelerators) onto a single chip. It enables powerful computing at the... links: [/agent/glossary/system-on-chip.md] ### [Tiny Machine Learning](/resources/glossary/tiny-machine-learning) - **Letter:** T - **Definition:** Tiny Machine Learning (TinyML) is the practice of deploying machine learning models directly on edge devices, enabling real-time intelligence without cloud dependency. Also called... links: [/agent/glossary/tiny-machine-learning.md] ### [Training](/resources/glossary/training) - **Letter:** T - **Definition:** Training is the process of feeding data to an AI model so it can learn patterns, optimize parameters, and improve decision-making accuracy. In Edge AI, this enables devices to... links: [/agent/glossary/training.md] ### [Training Split](/resources/glossary/training-split) - **Letter:** T - **Definition:** A training split is the process of dividing datasets into subsets for training, validation, and testing in machine learning, ensuring Edge AI models learn accurately and perform... links: [/agent/glossary/training-split.md] ### [Transfer Learning](/resources/glossary/transfer-learning) - **Letter:** T - **Definition:** Transfer Learning is a machine learning technique where a model developed for one task is adapted to perform a related task, reducing the need for large datasets. In Edge AI, it... links: [/agent/glossary/transfer-learning.md] ### [Transformation](/resources/glossary/transformation) - **Letter:** T - **Definition:** In Edge AI, transformation refers to the process of converting raw, decentralized data into actionable insights and automated decisions at the edge of the network. This digital... links: [/agent/glossary/transformation.md] ### [Unsupervised Learning](/resources/glossary/unsupervised-learning) - **Letter:** U - **Definition:** Unsupervised learning is a type of machine learning where Edge AI systems analyze and identify patterns in data without pre-labeled outcomes. Also known as self-guided learning,... links: [/agent/glossary/unsupervised-learning.md] ### [Validation Split](/resources/glossary/validation-split) - **Letter:** V - **Definition:** A validation split is the portion of a dataset reserved to evaluate an AI model’s performance during training. Also called a holdout set, it ensures that Edge AI systems can... links: [/agent/glossary/validation-split.md] ### [Virtual Sensor](/resources/glossary/virtual-sensor) - **Letter:** V - **Definition:** A virtual sensor, also known as a soft sensor, is a software-based system that estimates physical measurements using algorithms and existing sensor data. In Edge AI, virtual... links: [/agent/glossary/virtual-sensor.md] ### [Window](/resources/glossary/window) - **Letter:** W - **Definition:** A window in Edge AI is a defined timeframe or dataset segment used for real-time processing and analysis at the edge. Also called a data window, it enables devices to capture,... links: [/agent/glossary/window.md] ### [Wizard of Oz Prototype](/resources/glossary/wizard-of-oz-prototype) - **Letter:** W - **Definition:** A Wizard of Oz Prototype is an early-stage testing method where a human secretly performs tasks that a software or Edge AI system will eventually handle. Also called a simulated... links: [/agent/glossary/wizard-of-oz-prototype.md]