What does "Negative Transfer" mean?

Table of Contents

Negative transfer happens when learning from one task makes it harder to learn a different, new task. This can lead to worse results than if the new task was learned without any prior experience.

In many cases, especially in machine learning, this issue arises when the information from the first task does not relate well to the new one. Instead of helping, the knowledge gained can confuse or mislead the learner, resulting in poor performance.

To tackle negative transfer, researchers are developing methods to help systems adapt better when facing new tasks. These methods often include strategies that reset previous knowledge or focus on relevant information, aiming to improve the learning process and outcomes.