NNI provides state-of-the-art tuning algorithm in our builtin-assessors and makes them easy to use. Below is the brief overview of NNI current builtin Assessors:
NNI provides state-of-the-art tuning algorithm in our builtin-assessors and makes them easy to use. Below is the brief overview of NNI current builtin Assessors:
Note: Click the **Assessor's name** to get a detailed description of the algorithm, click the corresponding **Usage** to get the Assessor's installation requirements, suggested scenario and using example.
|Assessor|Brief Introduction of Algorithm|
|Assessor|Brief Introduction of Algorithm|
|---|---|
|---|---|
|**Medianstop**[(Usage)](#MedianStop)|Medianstop is a simple early stopping rule mentioned in the [paper](https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/46180.pdf). It stops a pending trial X at step S if the trial’s best objective value by step S is strictly worse than the median value of the running averages of all completed trials’ objectives reported up to step S.|
|[Medianstop](../src/sdk/pynni/nni/medianstop_assessor/README.md)[(Usage)](#MedianStop)|Medianstop is a simple early stopping rule. It stops a pending trial X at step S if the trial’s best objective value by step S is strictly worse than the median value of the running averages of all completed trials’ objectives reported up to step S.[Reference Paper](https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/46180.pdf)|
|[Curvefitting](https://github.com/Microsoft/nni/blob/master/src/sdk/pynni/nni/curvefitting_assessor/README.md)[(Usage)](#Curvefitting)|Curve Fitting Assessor is a LPA(learning, predicting, assessing) algorithm. It stops a pending trial X at step S if the prediction of final epoch's performance worse than the best final performance in the trial history. In this algorithm, we use 12 curves to fit the accuracy curve|
|[Curvefitting](../src/sdk/pynni/nni/curvefitting_assessor/README.md)[(Usage)](#Curvefitting)|Curve Fitting Assessor is a LPA(learning, predicting, assessing) algorithm. It stops a pending trial X at step S if the prediction of final epoch's performance worse than the best final performance in the trial history. In this algorithm, we use 12 curves to fit the accuracy curve. [Reference Paper](http://aad.informatik.uni-freiburg.de/papers/15-IJCAI-Extrapolation_of_Learning_Curves.pdf)|
NNI provides state-of-the-art tuning algorithm as our builtin-tuners and makes them easy to use. Below is the brief summary of NNI currently built-in Tuners:
NNI provides state-of-the-art tuning algorithm as our builtin-tuners and makes them easy to use. Below is the brief summary of NNI currently built-in Tuners:
Note: Click the **Tuner's name** to get a detailed description of the algorithm, click the corresponding **Usage** to get the Tuner's installation requirements, suggested scenario and using example.
|Tuner|Brief Introduction of Algorithm|
|Tuner|Brief Introduction of Algorithm|
|---|---|
|---|---|
|**TPE**[(Usage)](#TPE)|The Tree-structured Parzen Estimator (TPE) is a sequential model-based optimization (SMBO) approach. SMBO methods sequentially construct models to approximate the performance of hyperparameters based on historical measurements, and then subsequently choose new hyperparameters to test based on this model.|
|[TPE](../src/sdk/pynni/nni/hyperopt_tuner/README.md)[(Usage)](#TPE)|The Tree-structured Parzen Estimator (TPE) is a sequential model-based optimization (SMBO) approach. SMBO methods sequentially construct models to approximate the performance of hyperparameters based on historical measurements, and then subsequently choose new hyperparameters to test based on this model.[Reference Paper](https://papers.nips.cc/paper/4443-algorithms-for-hyper-parameter-optimization.pdf)|
|**Random Search**[(Usage)](#Random)|In Random Search for Hyper-Parameter Optimization show that Random Search might be surprisingly simple and effective. We suggest that we could use Random Search as the baseline when we have no knowledge about the prior distribution of hyper-parameters.|
|[Random Search](../src/sdk/pynni/nni/hyperopt_tuner/README.md)[(Usage)](#Random)|In Random Search for Hyper-Parameter Optimization show that Random Search might be surprisingly simple and effective. We suggest that we could use Random Search as the baseline when we have no knowledge about the prior distribution of hyper-parameters.[Reference Paper](http://www.jmlr.org/papers/volume13/bergstra12a/bergstra12a.pdf)|
|**Anneal**[(Usage)](#Anneal)|This simple annealing algorithm begins by sampling from the prior, but tends over time to sample from points closer and closer to the best ones observed. This algorithm is a simple variation on the random search that leverages smoothness in the response surface. The annealing rate is not adaptive.|
|[Anneal](../src/sdk/pynni/nni/hyperopt_tuner/README.md)[(Usage)](#Anneal)|This simple annealing algorithm begins by sampling from the prior, but tends over time to sample from points closer and closer to the best ones observed. This algorithm is a simple variation on the random search that leverages smoothness in the response surface. The annealing rate is not adaptive.|
|**Naive Evolution**[(Usage)](#Evolution)|Naive Evolution comes from Large-Scale Evolution of Image Classifiers. It randomly initializes a population-based on search space. For each generation, it chooses better ones and does some mutation (e.g., change a hyperparameter, add/remove one layer) on them to get the next generation. Naive Evolution requires many trials to works, but it's very simple and easy to expand new features.|
|[Naive Evolution](../src/sdk/pynni/nni/evolution_tuner/README.md)[(Usage)](#Evolution)|Naive Evolution comes from Large-Scale Evolution of Image Classifiers. It randomly initializes a population-based on search space. For each generation, it chooses better ones and does some mutation (e.g., change a hyperparameter, add/remove one layer) on them to get the next generation. Naive Evolution requires many trials to works, but it's very simple and easy to expand new features.[Reference paper](https://arxiv.org/pdf/1703.01041.pdf)|
|**SMAC**[(Usage)](#SMAC)|SMAC is based on Sequential Model-Based Optimization (SMBO). It adapts the most prominent previously used model class (Gaussian stochastic process models) and introduces the model class of random forests to SMBO, in order to handle categorical parameters. The SMAC supported by nni is a wrapper on the SMAC3 Github repo. Notice, SMAC need to be installed by `nnictl package` command.|
|[SMAC](../src/sdk/pynni/nni/smac_tuner/README.md)[(Usage)](#SMAC)|SMAC is based on Sequential Model-Based Optimization (SMBO). It adapts the most prominent previously used model class (Gaussian stochastic process models) and introduces the model class of random forests to SMBO, in order to handle categorical parameters. The SMAC supported by nni is a wrapper on the SMAC3 Github repo. Notice, SMAC need to be installed by `nnictl package` command.[Reference Paper,](https://www.cs.ubc.ca/~hutter/papers/10-TR-SMAC.pdf)[Github Repo](https://github.com/automl/SMAC3)|
|**Batch tuner**[(Usage)](#Batch)|Batch tuner allows users to simply provide several configurations (i.e., choices of hyper-parameters) for their trial code. After finishing all the configurations, the experiment is done. Batch tuner only supports the type choice in search space spec.|
|[Batch tuner](../src/sdk/pynni/nni/batch_tuner/README.md)[(Usage)](#Batch)|Batch tuner allows users to simply provide several configurations (i.e., choices of hyper-parameters) for their trial code. After finishing all the configurations, the experiment is done. Batch tuner only supports the type choice in search space spec.|
|**Grid Search**[(Usage)](#GridSearch)|Grid Search performs an exhaustive searching through a manually specified subset of the hyperparameter space defined in the searchspace file. Note that the only acceptable types of search space are choice, quniform, qloguniform. The number q in quniform and qloguniform has special meaning (different from the spec in search space spec). It means the number of values that will be sampled evenly from the range low and high.|
|[Grid Search](../src/sdk/pynni/nni/gridsearch_tuner/README.md)[(Usage)](#GridSearch)|Grid Search performs an exhaustive searching through a manually specified subset of the hyperparameter space defined in the searchspace file. Note that the only acceptable types of search space are choice, quniform, qloguniform. The number q in quniform and qloguniform has special meaning (different from the spec in search space spec). It means the number of values that will be sampled evenly from the range low and high.|
|[Hyperband](https://github.com/Microsoft/nni/tree/master/src/sdk/pynni/nni/hyperband_advisor)[(Usage)](#Hyperband)|Hyperband tries to use the limited resource to explore as many configurations as possible, and finds out the promising ones to get the final result. The basic idea is generating many configurations and to run them for the small number of STEPs to find out promising one, then further training those promising ones to select several more promising one.|
|[Hyperband](../src/sdk/pynni/nni/hyperband_advisor/README.md)[(Usage)](#Hyperband)|Hyperband tries to use the limited resource to explore as many configurations as possible, and finds out the promising ones to get the final result. The basic idea is generating many configurations and to run them for the small number of STEPs to find out promising one, then further training those promising ones to select several more promising one.[Reference Paper](https://arxiv.org/pdf/1603.06560.pdf)|
|[Network Morphism](https://github.com/Microsoft/nni/blob/master/src/sdk/pynni/nni/networkmorphism_tuner/README.md)[(Usage)](#NetworkMorphism)|Network Morphism provides functions to automatically search for architecture of deep learning models. Every child network inherits the knowledge from its parent network and morphs into diverse types of networks, including changes of depth, width, and skip-connection. Next, it estimates the value of a child network using the historic architecture and metric pairs. Then it selects the most promising one to train.|
|[Network Morphism](../src/sdk/pynni/nni/networkmorphism_tuner/README.md)[(Usage)](#NetworkMorphism)|Network Morphism provides functions to automatically search for architecture of deep learning models. Every child network inherits the knowledge from its parent network and morphs into diverse types of networks, including changes of depth, width, and skip-connection. Next, it estimates the value of a child network using the historic architecture and metric pairs. Then it selects the most promising one to train.[Reference Paper](https://arxiv.org/abs/1806.10282)|
|**Metis Tuner**[(Usage)](#MetisTuner)|Metis offers the following benefits when it comes to tuning parameters: While most tools only predict the optimal configuration, Metis gives you two outputs: (a) current prediction of optimal configuration, and (b) suggestion for the next trial. No more guesswork. While most tools assume training datasets do not have noisy data, Metis actually tells you if you need to re-sample a particular hyper-parameter.|
|[Metis Tuner](../src/sdk/pynni/nni/metis_tuner/README.md)[(Usage)](#MetisTuner)|Metis offers the following benefits when it comes to tuning parameters: While most tools only predict the optimal configuration, Metis gives you two outputs: (a) current prediction of optimal configuration, and (b) suggestion for the next trial. No more guesswork. While most tools assume training datasets do not have noisy data, Metis actually tells you if you need to re-sample a particular hyper-parameter.[Reference Paper](https://www.microsoft.com/en-us/research/publication/metis-robustly-tuning-tail-latencies-cloud-systems/)|
Batch tuner allows users to simply provide several configurations (i.e., choices of hyper-parameters) for their trial code. After finishing all the configurations, the experiment is done. Batch tuner only supports the type choice in [search space spec](../../../../../docs/SearchSpaceSpec.md).
Suggested sceanrio: If the configurations you want to try have been decided, you can list them in searchspace file (using choice) and run them using batch tuner.
Naive Evolution comes from [Large-Scale Evolution of Image Classifiers](https://arxiv.org/pdf/1703.01041.pdf). It randomly initializes a population based on search space. For each generation, it chooses better ones and do some mutation (e.g., change a hyperparameter, add/remove one layer) on them to get the next generation. Naive Evolution requires many trials to works, but it's very simple and easily to expand new features.
Grid Search performs an exhaustive searching through a manually specified subset of the hyperparameter space defined in the searchspace file. Note that the only acceptable types of search space are `choice`, `quniform`, `qloguniform`. **The number `q` in `quniform` and `qloguniform` has special meaning (different from the spec in [search space spec](../../../../../docs/SearchSpaceSpec.md)). It means the number of values that will be sampled evenly from the range `low` and `high`.**
The Tree-structured Parzen Estimator (TPE) is a sequential model-based optimization (SMBO) approach. SMBO methods sequentially construct models to approximate the performance of hyperparameters based on historical measurements, and then subsequently choose new hyperparameters to test based on this model. The TPE approach models P(x|y) and P(y) where x represents hyperparameters and y the associated evaluate matric. P(x|y) is modeled by transforming the generative process of hyperparameters, replacing the distributions of the configuration prior with non-parametric densities. This optimization approach is described in detail in [Algorithms for Hyper-Parameter Optimization](https://papers.nips.cc/paper/4443-algorithms-for-hyper-parameter-optimization.pdf).
## Random Search
In [Random Search for Hyper-Parameter Optimization](http://www.jmlr.org/papers/volume13/bergstra12a/bergstra12a.pdf) show that Random Search might be surprisingly simple and effective. We suggests that we could use Random Search as baseline when we have no knowledge about the prior distribution of hyper-parameters.
## Anneal
This simple annealing algorithm begins by sampling from the prior, but tends over time to sample from points closer and closer to the best ones observed. This algorithm is a simple variation on random search that leverages smoothness in the response surface. The annealing rate is not adaptive.
Medianstop is a simple early stopping rule mentioned in the [paper](https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/46180.pdf). It stops a pending trial X at step S if the trial’s best objective value by step S is strictly worse than the median value of the running averages of all completed trials’ objectives reported up to step S.
[Metis](https://www.microsoft.com/en-us/research/publication/metis-robustly-tuning-tail-latencies-cloud-systems/) offers the following benefits when it comes to tuning parameters: While most tools only predicts the optimal configuration, Metis gives you two outputs: (a) current prediction of optimal configuration, and (b) suggestion for the next trial. No more guess work!
While most tools assume training datasets do not have noisy data, Metis actually tells you if you need to re-sample a particular hyper-parameter.
While most tools have problems of being exploitation-heavy, Metis' search strategy balances exploration, exploitation, and (optional) re-sampling.
Metis belongs to the class of sequential model-based optimization (SMBO), and it is based on the Bayesian Optimization framework. To model the parameter-vs-performance space, Metis uses both Gaussian Process and GMM. Since each trial can impose a high time cost, Metis heavily trades inference computations with naive trial. At each iteration, Metis does two tasks:
It finds the global optimal point in the Gaussian Process space. This point represents the optimal configuration.
It identifies the next hyper-parameter candidate. This is achieved by inferring the potential information gain of exploration, exploitation, and re-sampling.
Note that the only acceptable types of search space are `choice`, `quniform`, `uniform` and `randint`.
More details can be found in our paper: https://www.microsoft.com/en-us/research/publication/metis-robustly-tuning-tail-latencies-cloud-systems/
[SMAC](https://www.cs.ubc.ca/~hutter/papers/10-TR-SMAC.pdf) is based on Sequential Model-Based Optimization (SMBO). It adapts the most prominent previously used model class (Gaussian stochastic process models) and introduces the model class of random forests to SMBO, in order to handle categorical parameters. The SMAC supported by nni is a wrapper on [the SMAC3 github repo](https://github.com/automl/SMAC3).
Note that SMAC on nni only supports a subset of the types in [search space spec](../../../../../docs/SearchSpaceSpec.md), including `choice`, `randint`, `uniform`, `loguniform`, `quniform(q=1)`.
