menu
announcement

Spectrum is now read-only. Learn more about the decision in our official announcement.

snorkel

Building and managing training datasets for machine learning

Channels
Team
Posts
Members
Info

Feedback on how to further improve a data labeling model using snorkel

Hi, I am working on a project to programmatically label messages as spam or ham (not spam) on this sms dataset (https://paperswithcode.com/dataset/sms-spam-collection-data-set). By referring to the snorkel github tutorial (which I found really helpful), I have been able to…

message-simple
0

Snorkel for Geospatial Data Activity Classification

Hi, I am currently working on the activity classification for points in a GPS trajectory, currently I want to classify each of the point in a GPS trajectory in one of the 3 activites based on the change in GPS direction(bearing difference) and the speed : 1. If the speed is low…

message-simple
6

I wrote a medium post on how to use snorkel for multi-label.

If you would like to use it, feel free. https://towardsdatascience.com/using-snorkel-for-multi-label-annotation-cc2aa217986a

message-simple
0

Is there going to be a dependency learning feature soon?

I wonder if there is a plan to integrate the dependency learning feature in the Snorkel?

message-simple
3

validate weak labels

I'm wondering if there are some papers that addressed how to validate weak labels without large-scale ground-truth labels?

message-simple
2

Definition of the label model

First of all, thanks for your great work on Snorkel, especially the new version. I have a question regarding the paper "Training complex models with multi-task weak supervision", which describes the new, matrix-style completion approach for estimating the label model in Snorkel.…

message-simple
2

Snorkel in Venture Capital

I do applied research at Georgian Partners, a growth-stage venture capital firm in Toronto, Canada. We have found Snorkel to be a useful tool in consolidating various heuristics on what makes a company successful / interesting for us. Furthermore, some of our portfolio companies…

message-simple
0

#MAGAZINEgts, a #DigitalHumanities #AI/#ML #CitizenScientist Research Project

I am a 68-year-old post-cancer #PayItForward #CitizenScientist working at the intersection of #DigitalHumanities and #AI/#ML. My applied research is focused on the development of #MAGAZINEgts, a ground-truth storage format for serial publications, especially historic magazines…

message-simple
0

New Member Self-Introductions, please

With this new channel for the Snorkel community to grow and communicate, as a new member I would very much like to know who we are and what our current research/activity interests are beyond the 140-character bios of our profiles. Thank you. P.S. Moderator/admin folk, perhaps…

message-simple
2