Maren Mahsereci

Hi, related to issue #121 this is the attempt at a one-vs-all sampler. The tasks created are binary classification tasks where class 0 represents one label and class 1 represents...

one-vs-all sampler - how to get binary targets?

2

Hi, great work on the library! I am attempting to implement a task sampler that samples 1-vs-all tasks. That is every task is a binary classification problem where the class...

`test_optimization_with_linear_constraint` in integration tests is flaky.

6

Test fails about 50% of the time due to infs or nans. `integration_tests/emukit/bayesian_optimization/test_constrained_loop.py::test_optimization_with_linear_constraint`

`ProductMatern12` kernel embedding w.r.t Lebesgue measure & GPy wrapper

1

*Issue #, if available:* *Description of changes:* This PR adds - kernel embeddings for the Matern12 (a.k.a. Exponential) product kernel w.r.t. Lebesgue measure. - an EmuKit interface for this kernel....

new release?

5

@apaleyes what do you think about a new release on pypi? The `quadrature` package changed quite a bit in the last months.

API references cleanup

4

The API references have several issues that should be addressed. These are, among others: - Some latex formulas are not rendered correctly - For some classes, the parameter descriptions are...

a tutorial notebook would be useful

clean up epmgp.py

2

it is not easy to read. Also we might want to move it out of the util folder

WSABI for BQ package

3

both WSABI-L and WSABI-M (Gunter et al. 2014) with their corresponding acquisition function might be a nice addition. - [ ] WSABI-L (is already useful without the variance implementation) -...

Stopping condition for cost sensitive evaluations

2

Currently the stopping condition is only checked at beginning of each loop-step. If the evaluation is cost/budget sensitive e.g., in multi-fidelity/source models or models where the input location determines the...