Caesar AI Atlas

Machine Learning Foundations

A foundations hub for the core machine-learning concepts that legal, compliance, and product teams need before reviewing AI systems.

beginnerslawyerscompliance teamsproduct managersresearchers

30

Terms

10

Comparisons

8

Questions

About This Topic

Overview

Machine learning foundations explain how models learn from data, how datasets are split, and why model behavior depends on training choices. This topic helps non-specialists understand enough technical context to ask better compliance questions.

Why This Matters

Many legal and governance disputes come from basic misunderstandings about training data, model behavior, validation, bias, and generalization.

Compliance Context

Useful for AI literacy, procurement review, model documentation, bias analysis, audit preparation, and cross-functional training.

Key Takeaways

  • 1Learning paradigm affects risk and evidence needs.
  • 2Training data, validation data, and test sets should not be confused.
  • 3Bias and fairness cannot be assessed only from model architecture.
  • 4Foundational terms improve communication between lawyers and engineers.

Key Terms

Learning paradigms

Core approaches that describe how systems learn from labels, data, feedback, or interaction.

Data and generalization

Dataset and training concepts that shape model quality and governance risk.

Related Comparisons

Technical Alternative

Supervised Learning vs Unsupervised Learning

A side-by-side comparison of Supervised Learning and Unsupervised Learning. Understand how learning from labeled examples differs from finding patterns in data without pre-existing labels.

Technical Alternative

Reinforcement Learning vs Supervised Learning

A side-by-side comparison of Reinforcement Learning and Supervised Learning. It explains how learning from rewards in an environment differs from learning mappings from labeled examples.

Technical Alternative

Supervised Learning vs Semi-Supervised Learning

A side-by-side comparison of Supervised Learning and Semi-Supervised Learning. Understand how learning from labeled examples differs from combining labeled and unlabeled data.

Technical Alternative

Self-supervised Learning vs Semi-Supervised Learning

A side-by-side comparison of Self-supervised Learning and Semi-Supervised Learning. Understand how each uses unlabeled data and why the role of labels differs.

Common Confusion

Machine Learning vs Deep Learning

A side-by-side comparison of Machine Learning and Deep Learning. Understand how the broader field of learning from data differs from neural-network methods with multiple layers.

Common Confusion

Training Data vs Test Set

A side-by-side comparison of Training Data and Test Set. Understand why the data used to fit a model must be separated from the reserved data used to evaluate it.

Common Confusion

Training Data vs Validation Data

A side-by-side comparison of Training Data and Validation Data. Understand how data used to fit the model differs from data used during development to tune choices and detect generalization problems.

Common Confusion

Overfitting vs Underfitting

A side-by-side comparison of Overfitting and Underfitting. It explains how learning training data too closely differs from failing to learn meaningful patterns in the data.

Common Confusion

Classification vs Regression

A side-by-side comparison of Classification and Regression. Understand how predicting discrete categories differs from predicting continuous or numerical values.

Technical Alternative

Decision Tree vs Random Forest

A side-by-side comparison of Decision Tree and Random Forest. Understand how the terms differ, when each applies, and what the distinction means for AI governance, system design, or assurance evidence.

Common Questions

What is self-supervised learning?

Self-supervised learning is a machine learning approach that creates training signals from unlabeled data itself. It converts aspects of an unsupervised problem into supervised learning by generating surrogate labels, making it useful when labeled data is scarce but unlabeled data is abundant.

1 related termsintermediate

What is semi-supervised learning?

Semi-supervised learning is a machine learning approach that uses both labeled and unlabeled data during training. It is useful when labeled examples are costly or limited but larger volumes of unlabeled data can help improve representation, classification, or prediction.

1 related termsintermediate

What is training data?

Training data is the dataset used to teach or fit a machine learning model by adjusting its learnable parameters. Its quality, quantity, diversity, labeling, and provenance strongly influence the performance, reliability, and compliance posture of the resulting model.

1 related termsintermediate

What is overfitting?

Overfitting occurs when a model learns the training data too closely, including noise or idiosyncratic details, and performs poorly on new data. It is commonly indicated by high training performance but weaker validation or test performance.

1 related termsintermediate

What is underfitting?

Underfitting occurs when a model is too simple or insufficiently trained to capture meaningful patterns in the data. An underfit model performs poorly on both training data and new data because it has not learned the underlying relationships needed for the task.

1 related termsintermediate

What is a decision tree?

A decision tree is an interpretable supervised learning model that represents decision rules in a tree-like structure. Inference follows a path from a root node through feature-based conditions to a leaf node that provides the predicted class or value.

1 related termsintermediate

What is deep learning?

Deep learning is a subfield of machine learning that uses neural networks with multiple layers to learn representations and patterns from data.

1 related termsadvanced

How are classification and regression different?

Use Classification when the target is a category; use Regression when the target is a continuous or numerical value.

1 related termsadvanced