tensorflow / tensorflow/probability
Possible improvement to `tfp.optimizer.bfgs_minimize`
Nobody has claimed this yet.
- Dominant language
- Jupyter Notebook
- Stars
- 4.4k
- Forks
- 1.1k
- PR merge metrics
- No merged PRs in 30d
Description
Hi,
I have some optimization problems which seem to be very sensitive to the choice of initial_inverse_hessian_estimate in tfp.optimizer.bfgs_minimize.
I am wondering if it is worth implementing the the algorithm described in Nocedal and Wright 6.20 to help with this? I would hope it isn't too big a change and might help some people.
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 with the tfp.optimizer.bfgs_minimize API documentation and compare its current behavior with the algorithm described in section 6.20 of Nocedal and Wright. Determine the required scope and validation cases for changing the initial inverse Hessian estimate, then confirm completion with optimization problems that demonstrate the intended improvement.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- tensorflow
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100