@@ -23,8 +23,11 @@ Assuming additive a Gaussian noise and the noise parameter is initialized to its
...
@@ -23,8 +23,11 @@ Assuming additive a Gaussian noise and the noise parameter is initialized to its
We determine the maximum probability value of the new combined parameter vector by learing the historical data. Use such value to predict the future trial performance, and stop the inadequate experiments to save computing resource.
We determine the maximum probability value of the new combined parameter vector by learing the historical data. Use such value to predict the future trial performance, and stop the inadequate experiments to save computing resource.
Concretely,this algorithm goes through three stages of learning, predicting and assessing.
Concretely,this algorithm goes through three stages of learning, predicting and assessing.
* Step1: Learning. We will learning about the trial history of the current trial and determine the \xi at Bayesian angle. First of all, We fit each curve using the least squares method(implement by `fit_theta`) to save our time. After we obtained the parameters, we filter the curve and remove the outliers(implement by `filter_curve`). Fially, we use the MCMC sampling method(implement by `mcmc_sampling`) to adjust the weight of each curve. Up to now, we have dertermined all the parameters in \xi.
* Step1: Learning. We will learning about the trial history of the current trial and determine the \xi at Bayesian angle. First of all, We fit each curve using the least squares method(implement by `fit_theta`) to save our time. After we obtained the parameters, we filter the curve and remove the outliers(implement by `filter_curve`). Finally, we use the MCMC sampling method(implement by `mcmc_sampling`) to adjust the weight of each curve. Up to now, we have dertermined all the parameters in \xi.
* Step2: Predicting. Calculates the expected final result accuracy(implement by `f_comb`) at target position(ie the total number of epoch) by the \xi and the formula of the combined model.
* Step2: Predicting. Calculates the expected final result accuracy(implement by `f_comb`) at target position(ie the total number of epoch) by the \xi and the formula of the combined model.
* Step3: If the fitting result doesn't converge, the predicted value will be `None`, in this case we return `AssessResult.Good` to ask for future accuracy information and predict again. Furthermore, we will get a positive value by `predict()` function, if this value is strictly greater than the best final performance in history *`THRESHOLD`(default value = 0.95), return `AssessResult.Good`, otherwise, return `AssessResult.Bad`
* Step3: If the fitting result doesn't converge, the predicted value will be `None`, in this case we return `AssessResult.Good` to ask for future accuracy information and predict again. Furthermore, we will get a positive value by `predict()` function, if this value is strictly greater than the best final performance in history *`THRESHOLD`(default value = 0.95), return `AssessResult.Good`, otherwise, return `AssessResult.Bad`
The figure below is the result of our algorithm on MNIST trial history data, where the green point represents the data obtained by Assessor, the blue point represents the future but unknown data, and the red line is the Curve predicted by the Curve fitting assessor.
The figure below is the result of our algorithm on MNIST trial history data, where the green point represents the data obtained by Assessor, the blue point represents the future but unknown data, and the red line is the Curve predicted by the Curve fitting assessor.
[Hyperband][1] is a popular automl algorithm. The basic idea of Hyperband is that it creates several brackets, each bracket has `n` randomly generated hyperparameter configurations, each configuration uses `r` resource (e.g., epoch number, batch number). After the `n` configurations is finished, it chooses top `n/eta` configurations and runs them using increased `r*eta` resource. At last, it chooses the best configuration it has found so far.
[Hyperband][1] is a popular automl algorithm. The basic idea of Hyperband is that it creates several buckets, each bucket has `n` randomly generated hyperparameter configurations, each configuration uses `r` resource (e.g., epoch number, batch number). After the `n` configurations is finished, it chooses top `n/eta` configurations and runs them using increased `r*eta` resource. At last, it chooses the best configuration it has found so far.
## 2. Implementation with fully parallelism
## 2. Implementation with fully parallelism
Frist, this is an example of how to write an automl algorithm based on MsgDispatcherBase, rather than Tuner and Assessor. Hyperband is implemented in this way because it integrates the functions of both Tuner and Assessor, thus, we call it advisor.
Frist, this is an example of how to write an automl algorithm based on MsgDispatcherBase, rather than Tuner and Assessor. Hyperband is implemented in this way because it integrates the functions of both Tuner and Assessor, thus, we call it advisor.
Second, this implementation fully leverages Hyperband's internal parallelism. More specifically, the next bracket is not started strictly after the current bracket, instead, it starts when there is available resource.
Second, this implementation fully leverages Hyperband's internal parallelism. More specifically, the next bucket is not started strictly after the current bucket, instead, it starts when there is available resource.
## 3. Usage
## 3. Usage
To use Hyperband, you should add the following spec in your experiment's yaml config file:
To use Hyperband, you should add the following spec in your experiment's yaml config file:
...
@@ -43,7 +43,7 @@ Here is a concrete example of `R=81` and `eta=3`:
...
@@ -43,7 +43,7 @@ Here is a concrete example of `R=81` and `eta=3`:
|3 |3 27 |1 81 | | | |
|3 |3 27 |1 81 | | | |
|4 |1 81 | | | | |
|4 |1 81 | | | | |
`s` means bracket, `n` means the number of configurations that are generated, the corresponding `r` means how many STEPS these configurations run. `i` means round, for example, bracket 4 has 5 rounds, bracket 3 has 4 rounds.
`s` means bucket, `n` means the number of configurations that are generated, the corresponding `r` means how many STEPS these configurations run. `i` means round, for example, bucket 4 has 5 rounds, bucket 3 has 4 rounds.
About how to write trial code, please refer to the instructions under `examples/trials/mnist-hyperband/`.
About how to write trial code, please refer to the instructions under `examples/trials/mnist-hyperband/`.
[Autokeras](https://arxiv.org/abs/1806.10282) is a popular automl tools using Network Morphism. The basic idea of Autokeras is to use Bayesian Regression to estimate the metric of the Neural Network Architecture. Each time, it generates several child networks from father networks. Then it uses a naïve Bayesian regression estimate its metric value from history trained results of network and metric value pair. Next, it chooses the the child which has best estimated performance and adds it to the training queue. Inspired by its work and referring to its [code](https://github.com/jhfjhfj1/autokeras), we implement our Network Morphism method in our NNI platform.
[Autokeras](https://arxiv.org/abs/1806.10282) is a popular automl tools using Network Morphism. The basic idea of Autokeras is to use Bayesian Regression to estimate the metric of the Neural Network Architecture. Each time, it generates several child networks from father networks. Then it uses a naïve Bayesian regression estimate its metric value from history trained results of network and metric value pair. Next, it chooses the the child which has best estimated performance and adds it to the training queue. Inspired by its work and referring to its [code](https://github.com/jhfjhfj1/autokeras), we implement our Network Morphism method in our NNI platform.
