Goptuna
Bayesian optimization library for black-box functions, inspired by Optuna. This library is not only for hyperparameter tuning of machine learning models but also we can use the parameter tuning of the systems like server middleware (e.g. Controlling the number of goroutines of your server) as much as we can design objective function.
Currently two algorithms are implemented:
- Random Search
- Tree-structured Parzen Estimators (TPE)
See the blog post for more details: Practical bayesian optimization in Go using Goptuna.
Installation
You can integrate Goptuna in wide variety of Go projects because of its portability of pure Go.
$ go get -u github.com/c-bata/goptunaUsage
Goptuna supports Define-By-Run style user API like Optuna. It makes the modularity high, and the user can dynamically construct the search spaces.
package main
import (
"fmt"
"math"
"github.com/c-bata/goptuna"
"github.com/c-bata/goptuna/tpe"
"go.uber.org/zap"
)
func objective(trial goptuna.Trial) (float64, error) {
x1, _ := trial.SuggestUniform("x1", -10, 10)
x2, _ := trial.SuggestUniform("x2", -10, 10)
return math.Pow(x1-2, 2) + math.Pow(x2+5, 2), nil
}
func main() {
logger, _ := zap.NewDevelopment()
defer logger.Sync()
study, _ := goptuna.CreateStudy(
"goptuna-example",
goptuna.StudyOptionSampler(tpe.NewSampler()),
goptuna.StudyOptionSetDirection(goptuna.StudyDirectionMinimize),
goptuna.StudyOptionSetLogger(logger),
)
_ = study.Optimize(objective, 100)
v, _ := study.GetBestValue()
params, _ := study.GetBestParams()
fmt.Println("result:", v, params)
}Advanced usages
Parallel optimization with multiple goroutine workers
Optimize method of goptuna.Study object is designed as the goroutine safe.
So you can easily optimize your objective function using multiple goroutine workers.
package main
import ...
func main() {
study, _ := goptuna.CreateStudy(...)
eg, ctx := errgroup.WithContext(context.Background())
study.WithContext(ctx)
for i := 0; i < 5; i++ {
eg.Go(func() error {
return study.Optimize(objective, 100)
})
}
if err := eg.Wait(); err != nil { ... }
...
}Receive notifications of each trials
You can receive notifications of each trials via channel. It can be used for logging and any notification systems.
package main
import ...
func main() {
trialchan := make(chan goptuna.FrozenTrial, 8)
study, _ := goptuna.CreateStudy(
...
goptuna.StudyOptionIgnoreObjectiveErr(true),
goptuna.StudyOptionSetTrialNotifyChannel(trialchan),
)
var wg sync.WaitGroup
wg.Add(2)
go func() {
defer wg.Done()
err = study.Optimize(objective, 100)
close(trialchan)
}()
go func() {
defer wg.Done()
for t := range trialchan {
log.Println("trial", t)
}
}()
wg.Wait()
if err != nil { ... }
...
}Links
References:
- TPE: James S. Bergstra, Remi Bardenet, Yoshua Bengio, and Balázs Kégl. Algorithms for hyper-parameter optimization. In Advances in Neural Information Processing Systems 25. 2011.
- Optuna: Takuya Akiba, Shotaro Sano, Toshihiko Yanase, Takeru Ohta, Masanori Koyama. 2019. Optuna: A Next-generation Hyperparameter Optimization Framework. In The 25th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’19), August 4–8, 2019.
Status:
License
This software is licensed under the MIT license, see LICENSE for more information.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent