tensorflow / tensorflow/privacy
PATE Accountant for Multi-label Classification
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 2k
- Forks
- 477
- Avg merge
- 22h 12m
- Merged PRs (30d)
- 1
Description
Hello community!
PATE's data-dependent analysis is based on teacher predictions. Is it possible to extend PATE to multi-label tasks such as part-of-speech tagging or named entity recognition, where the length of labels varies with sentence length? Fixed length sentences would also be fine.
Currently, each teacher outputs a nested list (fixed length sentences would also be fine).
Help or a hint that PATE is only possible for single-label classification would help me a lot.
Regards,
Stefan
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reviewing the existing PATE accountant entry point and how it handles teacher predictions. Compare those assumptions with nested, variable-length outputs for multi-label classification, including the fixed-length case mentioned in the issue. Done means establishing whether support is feasible and documenting the resulting behavior or limitation.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100