IBM Data Science Practice Test

The Watson Jeopardy! game utilized which type of machine learning?

1. Unsupervised

2. Supervised

3. Reinforcement

4. Semi-supervised

The correct answer is: Supervised

The Watson Jeopardy! game primarily utilized supervised machine learning. In the context of Watson, supervised learning involves training algorithms on a labeled dataset, where the input data and the corresponding output are known. For Watson to understand the nuances of language, make inferences, and respond to questions accurately in a quiz-like format, it relied heavily on vast amounts of structured data that had been previously labeled and categorized. This enabled the system to learn associations between different types of data, such as questions and their correct answers. It’s important to understand the nature of the task involved in Jeopardy! Watson was not simply performing a classification task or clustering data in an unsupervised manner, nor was it experimenting with its environment to improve over time as seen in reinforcement learning. Instead, it required a clear understanding of language provided through historical data in a supervised learning framework. Semi-supervised learning, while useful in scenarios where only part of the data is labeled, does not fit the primary architecture of how Watson processed information for the competitive game format. Thus, supervised learning is the most appropriate classification for the machine learning employed in Watson Jeopardy!