AIRiskAware
AI Governance Glossary
Governance Practice

What Is Transfer Learning?

Transfer Learning is a technique in which a model developed for one task is reused as the starting point for a model on a related task.

Definition

Transfer Learninga technique in which a model developed for one task is reused as the starting point for a model on a related task.

Transfer learning underpins most modern AI: rather than training from scratch, organisations take a model that has already learned general patterns from large datasets and adapt it to their specific use case (a process related to fine-tuning). The governance relevance is inheritance of risk — the adapted model inherits the biases, security weaknesses, and data-provenance questions of the base model, so due diligence on the upstream model becomes part of the deployer's own risk assessment.

Source: ISO/IEC 22989:2022

Plain-language explanation

Transfer learning underpins most modern AI: rather than training from scratch, organisations take a model that has already learned general patterns from large datasets and adapt it to their specific use case (a process related to fine-tuning). The governance relevance is inheritance of risk — the adapted model inherits the biases, security weaknesses, and data-provenance questions of the base model, so due diligence on the upstream model becomes part of the deployer's own risk assessment.

Primary source: ISO/IEC 22989:2022

See where you stand on AI governance

Take the free 7-question maturity assessment and get a personalised action plan.

Free assessment — 3 minutes →