Forager: An Interactive System for Rapid ML Model Creation
Abstract/Contents
- Abstract
- Applying machine learning (ML) in practice on real-world tasks and datasets is a difficult, expensive, and slow process, limiting the widespread adoption of ML. One reason applying ML is so difficult is because it is an iterative process that requires multiple types of expertise -- defining a category of interest, curating training data, labeling it, training a model, and reviewing results to determine necessary improvements requires both a subject matter expert (who knows the dataset and problem to be solved) and machine learning expert -- which limits the speed at which the iteration can occur. We propose Forager, a system that lets a subject matter expert interactively iterate within a tight loop to rapidly create an ML model themself; to accomplish this, Forager provides a set of tools for data exploration, manual data annotation, machine-aided data annotation, model training, and model validation. Using Forager, we were able to create image classification models for previously-unlabeled rare categories on the Waymo Open Dataset within tens of minutes.
Description
| Type of resource | text |
|---|---|
| Date created | May 9, 2021 |
Creators/Contributors
| Author | Garimella, Mihir | |
|---|---|---|
| Degree granting institution | Stanford University, Department of Computer Science | |
| Primary advisor | Fatahalian, Kayvon | |
| Advisor | Re, Christopher |
Subjects
| Subject | machine learning |
|---|---|
| Subject | active learning |
| Subject | interactive |
| Subject | machine learning infrastructure |
| Subject | ML infrastructure |
| Subject | ML |
| Subject | AI |
| Genre | Thesis |
Bibliographic information
Access conditions
- Use and reproduction
- User agrees that, where applicable, content will not be used to identify or to otherwise infringe the privacy or confidentiality rights of individuals. Content distributed via the Stanford Digital Repository may be subject to additional license and use restrictions applied by the depositor.
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Preferred citation
- Preferred Citation
- Garimella, Mihir. (2021). Forager: An Interactive System for Rapid ML Model Creation. Stanford Digital Repository. Available at: https://purl.stanford.edu/yk250gc0084
Collection
Undergraduate Theses, School of Engineering
View other items in this collection in SearchWorksContact information
- Contact
- engreference@stanford.edu
Also listed in
Loading usage metrics...