MLS-C01 Exam Prep
MLS-C01 Exam Glossary - 37 Terms
Search the terminology pack for AWS Certified Machine Learning - Specialty. Use these definitions with the study guide and practice questions.
A
- A/B testing
- An experimental method that compares two versions of a system by exposing different groups to each version and measuring outcomes.
- Amazon Transcribe
- An AWS managed service that converts speech audio into text for search, analysis, and downstream processing.
B
- business process improvement
- The practice of optimizing operational workflows, sometimes by applying machine learning when data-driven prediction is useful.
D
- data ingestion pipeline
- A workflow that collects and moves data from source systems into storage or processing systems.
- data preparation
- The process of cleaning, transforming, and validating data before modeling.
- data sufficiency
- The condition that enough relevant data is available to effectively train a machine learning model.
- dataset issue
- A problem in collected data, such as missing values, bias, noise, or inconsistency, that can reduce model quality.
- defined intervals
- Specified repeating time periods used to trigger scheduled jobs or workflows.
- deployment
- The process of making a trained machine learning model available for production use.
- deterministic rule
- A fixed and explicit rule that always produces the same output for the same input.
I
- inference
- The process of using a trained model to generate outputs from new input data.
J
- job scheduling
- The configuration of tasks to run automatically at defined times or recurring intervals.
L
- labeled data
- Data examples that include the correct output or target value for training supervised models.
M
- machine learning
- A method of building systems that learn patterns or relationships from data to make predictions or decisions.
- manual intervention
- Human action required to start, manage, or correct a process that could otherwise be automated.
- mitigation strategy
- A planned action to reduce the impact of a data or modeling problem before continuing development.
- ML model
- The trained artifact produced by machine learning that is used to generate predictions or decisions from input data.
- ML operations
- The practices and workflows used to deploy, monitor, maintain, and update machine learning systems in production.
- MLOps
- A discipline that combines machine learning, software engineering, and operations to manage the ML lifecycle.
- model performance
- A measure of how well a machine learning model achieves its intended objective.
- model refresh
- The update of a deployed model through retraining to maintain relevance or accuracy over time.
- model reliability
- The consistency and dependability of a model's behavior in real or expected conditions.
- model retraining
- The process of training an existing model again, usually with new data, to refresh or improve it.
P
- pattern recognition
- The identification of meaningful structures or relationships in data that a model can learn.
- prediction
- An output generated by a machine learning model based on input data.
R
- real-world conditions
- Production or live usage environments where model behavior is evaluated with actual users or operational data.
- recommendation model
- A machine learning model designed to suggest products, content, or actions to users.
- recurring execution
- The repeated automatic running of a process according to a schedule.
- retraining pipeline
- An automated sequence of steps used to retrain an existing machine learning model on updated or additional data.
S
- speech-to-text
- The automatic conversion of spoken audio into written text.
- subject matter expert
- A person with deep domain knowledge who can define clear rules or provide guidance for a task.
- supervised learning
- A machine learning approach that learns from examples paired with correct target values.
T
- target value
- The correct output associated with an input example in supervised learning.
- trained artifact
- The resulting learned object from model training that can be deployed for inference.
- training data
- The dataset used to teach a machine learning model during the training process.
- transcript
- The text output created from spoken audio by a speech recognition system.
W
- workflow automation
- The use of scheduled or event-driven mechanisms to run processes without manual intervention.
About These Definitions
These definitions are loaded from the shared release pack. Use them with the study guide and practice questions to connect vocabulary to exam scenarios.