For the Stanford community.
Is a lack of training data a major bottleneck for your machine learning applications?
Do you have mountains of potentially interesting data buried in PDFs, charts, websites, or unstructured text reports?
Is it difficult to use your domain knowledge to train useful machine learning models faster?
If so, the Hazy Research team from Stanford University invites you to attend the 2019 Snorkel Workshop on June 24-25, 2019. We will work with stakeholders from academia, industry, and government to demonstrate how the open-source Snorkel software system can be used to leverage weaker, noisier forms of supervision to train high-performance machine learning models in a fraction of the time that labeling data would require. This award-winning approach has recently been deployed in core products at the world’s largest technology companies, including Google; in the fight against human trafficking under DARPA’s MEMEX project; and in support of real medical applications in imaging and genomics.
One Day 1, members of the Snorkel research team will guide participants through an introduction to Snorkel, comprising lecture components interspersed with a hands-on demonstration of application to a relation extraction task over unstructured text. On Day 2, we will present extensions to Snorkel that handle richly formatted data (e.g PDFs, websites, etc.), non-text data types (e.g. images), and multiple tasks (e.g. the GLUE benchmark), provide support while users explore hands-on demos for each of these extensions, and engage participants in developing plans for their own Snorkel applications.
The ideal participant will be comfortable with Python, and will have visibility on how Snorkel could be deployed to support applications within their organization or research group. Funding may be available to support travel. Please direct questions to Jared Dunnmon. Please register interest in attending by filling out the intake form at the following link.