CertNexus CET key terms

A person or group that carries out malicious activity against systems or data, motivated by money, ideology, espionage, or disruption. Distinguished from accidental insider mistakes, which cause harm without intent.

The total set of points where an attacker could try to enter or extract data from a system, including network services, user accounts, applications, and physical devices. Shrinking it reduces exposure.

A social-engineering attack that impersonates a trusted party through email, messaging, or a fake login page to trick a person into revealing credentials or running malicious code.

Malicious software that encrypts or steals data and demands payment for release or non-disclosure. Recovery depends on tested backups and containment, not on paying the attacker.

An access control method requiring two or more independent evidence types, such as a password plus a hardware token or biometric. It blocks most credential-replay attacks even when a password leaks.

A design principle granting each user, service, or process only the permissions needed for its current task, for only as long as needed. It limits the damage a compromised account can cause.

A security model that treats every request as untrusted until verified, regardless of network location. Access decisions rely on continuous identity, device-health, and context checks rather than a trusted internal perimeter.

A documented, rehearsed sequence of roles and steps for detecting, containing, eradicating, and recovering from a security incident, plus the lessons-learned review that follows.

Manipulating people into breaking normal security procedures by exploiting trust, urgency, authority, or curiosity. Technical controls alone cannot stop it, so awareness training and verification habits matter.

The legal and contractual duty to inform regulators and affected individuals within set timeframes after personal data is exposed. Requirements vary by jurisdiction and by the data type involved.

Processing data near where it is generated, such as on a gateway or device, instead of sending everything to a distant data center. It lowers latency and bandwidth use but spreads security responsibility.

An intermediary device that aggregates traffic from many sensors and translates between local protocols and upstream networks. It often enforces filtering, authentication, and local data reduction.

A connected device with limited processing power, memory, energy, or bandwidth, such as a battery sensor. These limits rule out heavy encryption or patching approaches used on servers.

The process of enrolling a new device into a network: establishing identity, installing credentials and configuration, and registering it with management systems. Weak provisioning is a common source of default-password exposures.

Username and password values embedded in firmware or software that cannot be changed by the owner. They are a frequent IoT weakness because attackers can extract them from one device and reuse them widely.

Delivering firmware fixes to devices remotely, ideally with signed packages and rollback support. It is the main way to remediate vulnerabilities at scale, but it can brick devices if verification fails.

Computing and control systems that monitor or drive physical processes, such as industrial controllers and building systems. Availability and safety usually outrank confidentiality, so IT security practices need adaptation.

A software model kept in sync with a real asset using sensor data, used for simulation, monitoring, and predictive maintenance. It concentrates operational data, so access control and data integrity are critical.

Training a model on labeled examples so it learns to map inputs to known outputs. It covers classification and regression tasks and requires representative, correctly labeled data.

Finding structure in unlabeled data, for example by grouping similar records or reducing dimensions. There is no ground-truth answer to check against, so results need human interpretation.

Records that pair an input with the correct answer, such as an image and its category. Labels are expensive to produce and their quality caps what a supervised model can learn.

When a model memorizes training examples, including their noise, and performs poorly on new data. It is detected by a widening gap between training and validation performance.

Measuring a trained model's performance on data it did not learn from, using metrics suited to the task such as precision, recall, or error rate. It estimates how the model will behave in production.

A table comparing predicted versus actual classes, showing true and false positives and negatives. From it, precision, recall, and related metrics are computed to judge a classifier's trade-offs.

Identifying records or events that deviate markedly from expected patterns, often without labeled examples of the anomaly. It supports fraud alerts, fault monitoring, and security telemetry analysis.

Grouping unlabeled records so that items within a group are more similar to each other than to items outside it. Common uses include customer segmentation and exploratory pattern discovery.

A gradual change in real-world input data that no longer matches what a model was trained on, degrading its accuracy. Monitoring inputs and retraining are the usual responses.

The human or assisted process of attaching labels, tags, or boundaries to raw data so it can train or evaluate models. Clear guidelines and inter-annotator checks protect label quality.

The degree to which data is accurate, complete, consistent, timely, and fit for its intended use. Poor quality undermines both analytics and model training regardless of technique.

A record of where data came from, how it moved, and what transformations it underwent. It supports debugging, audit, impact analysis, and reproducibility of analytical results.

Detecting and correcting or removing inaccurate, duplicated, malformed, or missing values before analysis. Every correction should be logged so results stay reproducible and auditable.

Summarizing and visualizing a dataset before modeling to reveal distributions, outliers, relationships, and gaps. It guides which cleaning steps and methods are appropriate.

Correlation means two variables move together; causation means one actually produces change in the other. Confounders can create correlation, so observational findings need careful design before causal claims.

A distortion that occurs when the collected data does not represent the population being studied, because some groups are more likely to be included. Conclusions then fail to generalize.

Data that can identify a specific individual directly or when combined with other data, such as names, identifiers, or precise location. It triggers privacy obligations wherever it is stored or shared.

Collecting and retaining only the personal data actually needed for a stated purpose, and for no longer than necessary. It reduces breach impact and simplifies compliance.

Permission given by a person after receiving clear, understandable information about what data will be collected, why, how it will be used, and what choices they have.

The principle that personal data collected for one stated purpose should not be repurposed for unrelated uses without a lawful basis or fresh consent.

Irreversibly altering data so no individual can be re-identified, even by combining datasets. Weak techniques such as simple de-identification often fail against linkage attacks.

Systematic unfair outcomes produced by a model or automated decision system, often traceable to biased training data, flawed labels, or proxies for protected traits.

The degree to which a person can understand why a model produced a given output. It matters for contestability, safety, and regulatory accountability in high-stakes decisions.

The policies, roles, and controls that define who may access, change, share, and retain data, and how those rules are enforced and audited across an organization.

A structured review, done before a high-risk processing activity, that identifies privacy risks to individuals and the safeguards that will reduce them. It documents accountability decisions.