training data
Model's teaching material
Training data is the information used to teach a machine learning system.
The model learns patterns from this data—text, images, audio, or other inputs—that it then applies to new situations.
The quality and scope of training data largely determine what the system can do: The chatbot's training data included millions of books and websites.
Concerns about training data include copyright, bias, and whether the data reflects the real world adequately.
- datasetOrganized collection of information
- corpusAssembled written works
- held outMaintained resistance or position
- samplesSmall test portions
- labelsIdentifying markers attached to items
annotations · machine learning · model · neural network · overfitting · preprocessing · features · validation · testing · supervised · examples · curated · scraped · web scraping · synthetic · bias · cleaning · augmentation
- machine learningneural networks, deep learning, backpropagation, underfitting, CNNs
- neural networksneural network, neural networks, hidden layers, backpropagation, layers
- corpus linguisticscorpus, word counts, word frequency, concordance, n-grams
- AIAI, machine learning, deep learning, algorithm, chatbot
- Datadatasets, databank, microdata, measurements, statistics
- data sciencealgorithm, machine learning, dataset, model, analytics