General Questions¶
What is Machine Teaching?¶
Machine teaching is an iterative and interactive process that enables humans to teach skills to machines the same way they teach other humans. The process involves the human to build an AI model for problems where labeled data is not readily available. Machine teaching enables the human to communicate their domain knowledge in the form of labels and features. With every action the human performs (e.g. add a label or a feature), the model is trained and the system gives feedback on the best next action to maximize the model accuracy.
What is the difference between Machine Learning and Machine Teaching?¶
Machine learning is about extracting knowledge from data when plenty of labeled data is readily available. Machine teaching is about enabling humans to communicate their domain knowledge to the machine for problems where enough labeled data is not readily available.
What is Duet?¶
Duet is a machine teaching platform for text analysis. It enables developers and data analysts who have no machine learning expertise to build text analysis models namely document classifiers and entity extractors starting at zero labeled data. For example, you can categorize products based on their product description or classify product reviews based on the review sentiment. The Duet platform has two major components: namely, guiding the user to build a high quality custom AI model and managing the lifecycle of the model through MLOps.
Models in Duet categorize text based on a user-defined schema.
Much of the documentation and terminology in our platform will refer to classification models as classifiers, and to extraction models as extractors. Duet only offers classifiers and entity extractors will be offered soon.
Learn more about Duet's competitive advantage.
What is the difference between Intelus and Duet?¶
Intelus is the company and Duet is its first platform that enables users to build AI models to process their text streams.
Is my data safe?¶
Data privacy is very important to us. All data in Duet is secured both in transit and at rest. User data is not used in any experimentation to improve the service. Duet is undergoing the SOC2 compliance process and it should be completed soon.
Can I use Duet if I don't know how to program?¶
Yes. Duet offers a full no-code experience to build and operationalize a machine learning model. You iterate to teach the model incrementally with guidance from Duet. Then you deploy the model with few button clicks and integrate it with GoogleSheets or Zendesk to apply the model on your data.
Do I need to have labeled data in order to use Duet?¶
Duet expects unlabeled data to build a model. Duet will then guide you through the teaching process to communicate your domain expertise to the system while building the model.
What formats of unlabeled data does Duet accept?¶
You can upload unlabeled datasets in either CSV file or zipped file format. You can also upload data from Zendesk, Google Sheets and Reddit. You can learn more about importing data here.
What languages does Duet support?¶
Duet supports English at this time. Please contact us at support@intelus.ai to find out more information on our language roadmap.
Where can I receive support?¶
You can email support@intelus.ai.
What is the Duet pricing model?¶
Duet is given for free until mid Jan. and the pricing model will have 3 tiers free, standard and entperprise.
