Case Studies > Georgetown University's CSET Accelerates AI Development with Snorkel Flow

Georgetown University's CSET Accelerates AI Development with Snorkel Flow

Company Size
1,000+
Region
  • America
Country
  • United States
Product
  • Snorkel Flow
Tech Stack
  • NLP
  • Active Learning
Implementation Scale
  • Enterprise-wide Deployment
Impact Metrics
  • Customer Satisfaction
  • Digital Expertise
  • Productivity Improvements
Technology Category
  • Analytics & Modeling - Machine Learning
  • Analytics & Modeling - Natural Language Processing (NLP)
Applicable Industries
  • Education
  • Software
Applicable Functions
  • Product Research & Development
  • Quality Assurance
Use Cases
  • Automated Disease Diagnosis
  • Predictive Quality Analytics
Services
  • Data Science Services
  • Software Design & Engineering Services
About The Customer
Georgetown University's Center for Security and Emerging Technology (CSET) is a leading research organization focused on studying the security implications of emerging technologies. The center aims to provide policymakers with data-driven analysis and recommendations to navigate the complex landscape of technological advancements. CSET's data science team is dedicated to developing advanced AI models to classify and analyze scientific articles, particularly in the field of virology. The team faced significant challenges in managing the manual labeling workflows required for their NLP applications, which led them to seek more efficient and scalable solutions.
The Challenge
CSET's data science team wanted to build NLP applications that classify scientific articles such as technical papers on virology. The team realized that manual labeling workflows would be impractical for the task.
The Solution
With the help of integrated analysis tools within Snorkel Flow, the team was able to pinpoint data slices for domain expert spot-checks and troubleshooting to improve accuracy. This approach powered an active learning workflow, significantly enhancing the efficiency of their labeling process. Snorkel Flow's advanced features, such as autosuggest and cluster labeling functions (LFs), enabled the team to create 107,000 programmatic labels. This not only reduced the labeling time by 50% but also improved productivity and accuracy. Within days, the team achieved 85% accuracy on a classification model, demonstrating the effectiveness of the solution.
Operational Impact
  • The team was able to create 107,000 programmatic labels using Snorkel Flow's advanced features like autosuggest and cluster LFs.
  • There was a 50% reduction in labeling time, which significantly improved the team's productivity.
  • The active learning workflow allowed for pinpointing data slices for domain expert spot-checks and troubleshooting, enhancing the accuracy of the models.
  • The solution enabled the team to achieve 85% accuracy on a classification model within days, showcasing the rapid improvement in model performance.
Quantitative Benefit
  • 107,000 programmatic labels created.
  • 50% reduction in labeling time.
  • 85% accuracy on a classification model within days.

Case Study missing?

Start adding your own!

Register with your work email and create a new case study profile for your business.

Add New Record

Related Case Studies.

Contact us

Let's talk!
* Required
* Required
* Required
* Invalid email address
By submitting this form, you agree that IoT ONE may contact you with insights and marketing messaging.
No thanks, I don't want to receive any marketing emails from IoT ONE.
Submit

Thank you for your message!
We will contact you soon.