Active Learning boost for your ML problem
Successful artificial intelligence solutions always require a massive amount of high-quality labeled data. Unfortunately, in most cases, we don’t have a large and qualitative labeled set together. Even in such a big company as Wix, with a whole in-house labeling department, we still experience challenges with acquiring stunning datasets.
As a Data Scientist, I am using many tools to mitigate data issues, and one of them is Active Learning. At Wix.com, we use it for text classification (business categorization) problems, sentiment analysis on user support requests, objects, and portraits segmentation.
In my post, I described how to apply the Active Learning tool and customize it due to your needs for leveraging the success in your DS project. If you have limited resources, domain-specific data, an imbalanced dataset, or want to speed up your DS project creation — give a chance to Active Learning!
You can review the post here. Also, take a look at the video at FWdays conference about Active Learning and Weak Supervision usage to solve NLP problems.
