实验经理

设计和运行实验以训练和比较深度学习网络

expand all in page

Description

这实验经理应用程序使您能够创建深度学习实验，以在多个初始条件下培训网络并比较结果。例如，您可以使用深度学习实验来：

Sweep through a range of hyperparameter values or use Bayesian optimization to find optimal training options. Bayesian optimization requires Statistics and Machine Learning Toolbox™.
使用内置功能trainNetwork或者define your own custom training function.
Compare the results of using different data sets or test different deep network architectures.

要快速设置实验，您可以从预配置模板开始。实验模板支持包括图像分类，图像回归，序列分类，语义万博1manbetx分割和自定义训练循环的工作流程。

实验管理器提供可视化工具，例如训练图和混乱矩阵，过滤器以完善您的实验结果以及注释以记录您的观察结果。为了提高可重复性，每次运行实验时，实验经理都会存储实验定义的副本。您可以访问过去的实验定义，以跟踪产生每个结果的超参数组合。

实验经理组织您的实验，并导致project。

您可以在同一项目中存储多个实验。
Each experiment contains a set ofresults每次运行实验。
Each set of results consists of one or more试验corresponding to a different combination of hyperparameters.

默认情况下，实验经理一次进行一次试验。If you have Parallel Computing Toolbox™, you can configure your experiment to run multiple trials at the same time or to run a single trial at a time on multiple GPUs, on a cluster, or in the cloud. If you haveMATLAB^®Parallel Server™，您还可以将实验作为远程集群中的批处理作业卸载，以便您可以在培训期间继续工作或关闭MATLAB会话。有关更多信息，请参阅使用实验经理并行培训网络和Offload Experiments as Batch Jobs to Cluster。

这实验浏览器pane displays the hierarchy of experiments and results in the project. For instance, this project has two experiments, each of which has several sets of results.

这blue flaskindicates a built-in training experiment that uses thetrainNetwork功能。绿色烧杯indicates a custom training experiment that relies on a different training function. To open the configuration for an experiment and view its results, double-click the name of the experiment or a set of results.

打开实验经理应用程序

MATLAB Toolstrip: On theApps标签，下机器学习和深度学习，单击应用程序图标。
MATLAB命令提示：输入实验管理器。

Examples

expand all

Image Classification by Sweeping Hyperparameters

Open Example

这个例子展示了如何使用实验template for image classification by sweeping hyperparameters. With this template, you can quickly set up a built-in training experiment that uses thetrainNetwork功能。为了more examples of solving image classification problems with Experiment Manager, seeCreate a Deep Learning Experiment for Classification和使用实验经理并行培训网络。为了more information on an alternative strategy to sweeping hyperparameters, seeTune Experiment Hyperparameters by Using Bayesian Optimization。

Open the example to load a project with a preconfigured experiment that you can inspect and run. To open the experiment, in the实验浏览器pane, double-click the name of the experiment (实验1).

另外，您可以按照以下步骤配置实验。

1. Open Experiment Manager. A dialog box provides links to the getting started tutorials and your recent projects, as well as buttons to create a new project or open an example from the documentation.

2. Under新的，，，，selectProject。A dialog box lists several templates that support workflows including image classification, image regression, sequence classification, semantic segmentation, and custom training loops.

3. UnderImage Classification Experiments，，，，selectImage Classification by Sweeping Hyperparameters。

4.指定新项目的名称和位置。实验经理开设了该项目的新实验。这实验pane displays the description, hyperparameters, setup function, and metrics that define the experiment.

5. In theDescriptionfield, enter a description of the experiment:

Classification of digits, using various initial learning rates.

6. UnderHyperparameters，，，，replace the value ofmyInitialLearnRatewith0.0025:0.0025:0.015。Verify thatStrategyis set toExhaustive Sweep。

7. UnderSetup Function，，，，clickEdit。这setup function opens in MATLAB Editor. The setup function specifies the training data, network architecture, and training options for the experiment. In this experiment, the setup function has three sections.

负载培训数据defines image datastores containing the training and validation data for the experiment. The experiment uses the Digits data set, which consists of 10,000 28-by-28 pixel grayscale images of digits from 0 to 9, categorized by the digit they represent. For more information on this data set, seeImage Data Sets。
定义网络体系结构defines the architecture for a simple convolutional neural network for deep learning classification.
指定培训选项defines a训练实验对象。在这个实验中,setup function loads the values for the initial learning rate from themyInitialLearnRate在超参数表中输入。

When you run the experiment, Experiment Manager trains the network defined by the setup function six times. Each trial uses one of the learning rates specified in the hyperparameter table. By default, Experiment Manager runs one trial at a time. If you have Parallel Computing Toolbox, you can run multiple trials at the same time or offload your experiment as a batch job in a cluster.

一次进行实验的一个试验，在实验管理器工具条上模式，，，，select顺序和clickRun。
To run multiple trials at the same time, under模式，，，，select同时和clickRun。If there is no current parallel pool, Experiment Manager starts one using the default cluster profile. Experiment Manager then executes multiple simultaneous trials, depending on the number of parallel workers available. For best results, before you run your experiment, start a parallel pool with as many workers as GPUs. For more information, see使用实验经理并行培训网络和GPU Support by Release(Parallel Computing Toolbox)。
To offload the experiment as a batch job, under模式，，，，selectBatch Sequential或者批量同时，指定您的Cluster和Pool Size，然后单击Run。有关更多信息，请参阅Offload Experiments as Batch Jobs to Cluster。

A table of results displays the accuracy and loss for each trial.

实验运行时，请单击Training Plotto display the training plot and track the progress of each trial. You can also monitor the training progress in the MATLAB Command Window.

Click混淆矩阵在每个完成的试验中显示验证数据的混淆矩阵。

实验完成后，您可以使用列或过滤器试验对表进行排序Filters窗格。您还可以通过在结果表中添加注释来记录观察值。有关更多信息，请参阅Sort, Filter, and Annotate Experiment Results。

要测试单个试验的性能，请导出训练有素的网络或试验的培训信息。在实验经理工具条，选择Export>训练有素的网络或者Export>培训信息，，，，respectively. For more information, see网和信息。To save the contents of the results table as a桌子在MATLAB工作区中的数组，选择Export>Results Table。

To close the experiment, in the实验浏览器pane, right-click the name of the project and selectClose Project。实验经理closes all of the experiments and results contained in the project.

图像回归通过清除超参数

Open Example

这个例子展示了如何使用实验template for image regression by sweeping hyperparameters. With this template, you can quickly set up a built-in training experiment that uses thetrainNetwork功能。有关解决实验经理回归问题的另一个示例，请参见Create a Deep Learning Experiment for Regression。为了more information on an alternative strategy to sweeping hyperparameters, seeTune Experiment Hyperparameters by Using Bayesian Optimization。

另外，您可以按照以下步骤配置实验。

3. UnderImage Regression Experiments，，，，select图像回归通过清除超参数。

4.指定新项目的名称和位置。实验经理开设了该项目的新实验。这实验pane displays the description, hyperparameters, setup function, and metrics that define the experiment.

5. In theDescriptionfield, enter a description of the experiment:

Regression to predict angles of rotation of digits, using various initial learning rates.

6. UnderHyperparameters，，，，replace the value ofmyInitialLearnRatewith0.001:0.001:0.006。Verify thatStrategyis set toExhaustive Sweep。

负载培训数据将实验的培训和验证数据定义为4-D阵列。培训和验证数据每个由数字数据集中的5000张图像组成。每个图像显示一个从0到9的数字，以一定角度旋转。回归值对应于旋转角度。有关此数据集的更多信息，请参阅Image Data Sets。
定义网络体系结构defines the architecture for a simple convolutional neural network for deep learning regression.
指定培训选项defines a训练实验对象。在这个实验中,setup function loads the values for the initial learning rate from themyInitialLearnRate在超参数表中输入。

一次进行实验的一个试验，在实验管理器工具条上模式，，，，select顺序和clickRun。
To run multiple trials at the same time, under模式，，，，select同时和clickRun。If there is no current parallel pool, Experiment Manager starts one using the default cluster profile. Experiment Manager then executes multiple simultaneous trials, depending on the number of parallel workers available. For best results, before you run your experiment, start a parallel pool with as many workers as GPUs. For more information, see使用实验经理并行培训网络和GPU Support by Release(Parallel Computing Toolbox)。
To offload the experiment as a batch job, under模式，，，，selectBatch Sequential或者批量同时，指定您的Cluster和Pool Size，然后单击Run。有关更多信息，请参阅Offload Experiments as Batch Jobs to Cluster。

A table of results displays the root mean squared error (RMSE) and loss for each trial.

实验运行时，请单击Training Plotto display the training plot and track the progress of each trial. You can also monitor the training progress in the MATLAB Command Window.

To close the experiment, in the实验浏览器pane, right-click the name of the project and selectClose Project。实验经理closes all of the experiments and results contained in the project.

Image Classification Using Custom Training Loop

Open Example

This example shows how to use the training experiment template for image classification using a custom training loop. With this template, you can quickly set up a custom training experiment.

另外，您可以按照以下步骤配置实验。

3. UnderImage Classification Experiments，，，，selectImage Classification Using Custom Training Loop。

4. Select the location and name for a new project. Experiment Manager opens a new experiment in the project. The实验pane displays the description, hyperparameters, and training function that define the experiment.

3. In theDescriptionfield, enter a description of the experiment:

Classification of digits, using various initial learning rates.

4. UnderHyperparameters，，，，replace the value ofmyInitialLearnRatewith0.0025:0.0025:0.015。Verify thatStrategyis set toExhaustive Sweep。

5.下Training Function，，，，clickEdit。这training function opens in MATLAB Editor. The training function specifies the training data, network architecture, training options, and training procedure used by the experiment. In this experiment, the training function has four sections.

负载培训数据defines the training data for the experiment as 4-D arrays. The experiment uses the Digits data set, which consists of 5,000 28-by-28 pixel grayscale images of digits from 0 to 9, categorized by the digit they represent. For more information on this data set, seeImage Data Sets。
定义网络体系结构defines the architecture for a simple convolutional neural network for deep learning classification. To train the network with a custom training loop, the training function represents the network as adlnetwork目的。
指定培训选项defines the training options used by the experiment. In this experiment, the training function loads the values for the initial learning rate from themyInitialLearnRate在超参数表中输入。
火车模型定义实验使用的自定义训练循环。对于每个时期，自定义训练循环都会将数据调整并迭代微型数据。对于每个迷你批次，自定义培训循环评估模型梯度，状态和损失，确定基于时间的衰减学习率计划的学习率，并更新网络参数。为了跟踪训练的进度并记录训练损失的价值，培训功能使用experiments.Monitor目的monitor。

当您运行实验时，实验经理会训练由训练功能定义的网络六次。每个试验都使用超参数表中指定的学习率之一。默认情况下，实验经理一次进行一次试验。如果您具有并行计算工具箱，则可以同时运行多个试验，也可以将实验作为群集中的批处理作业。

一次进行实验的一个试验，在实验管理器工具条上模式，，，，select顺序和clickRun。
To run multiple trials at the same time, under模式，，，，select同时和clickRun。If there is no current parallel pool, Experiment Manager starts one using the default cluster profile. Experiment Manager then executes multiple simultaneous trials, depending on the number of parallel workers available. For best results, before you run your experiment, start a parallel pool with as many workers as GPUs. For more information, see使用实验经理并行培训网络和GPU Support by Release(Parallel Computing Toolbox)。
To offload the experiment as a batch job, under模式，，，，selectBatch Sequential或者批量同时，指定您的Cluster和Pool Size，然后单击Run。有关更多信息，请参阅Offload Experiments as Batch Jobs to Cluster。

A table of results displays the training loss for each trial.

实验运行时，请单击Training Plotto display the training plot and track the progress of each trial.

To test the performance of an individual trial, export the training output for the trial. On the实验经理工具条，选择Export>Training Output。在这个实验中,training output is a structure that contains the values of the training loss and the trained network. To save the contents of the results table as a桌子在MATLAB工作区中的数组，选择Export>Results Table。

To close the experiment, in the实验浏览器pane, right-click the name of the project and selectClose Project。实验经理closes all of the experiments and results contained in the project.

Configure Built-In Training Experiment

This example shows how to set up a built-in training experiment using the Experiment Manager app. Built-in training experiments rely on thetrainNetwork功能和support workflows such as image classification, image regression, sequence classification, and semantic segmentation.

Built-in training experiments consist of a description, a table of hyperparameters, a setup function, and a collection of metric functions to evaluate the results of the experiment.

在里面Descriptionfield, enter a description of the experiment.

UnderHyperparameters，，，，select the strategy to use for your experiment.

To sweep through a range of hyperparameter values, setStrategytoExhaustive Sweep。在里面hyperparameter table, enter the names and values of the hyperparameters used in the experiment. Hyperparameter names must start with a letter, followed by letters, digits, or underscores. Hyperparameter values must be scalars or vectors with numeric, logical, or string values, or cell arrays of character vectors. For example, these are valid hyperparameter specifications:
- 0.01
- 0.01：0.01：0.05
- (0.010.02 0.04 0.08]
- ("sgdm" "rmsprop" "adam"]
- {'squeezenet' 'googlenet' 'resnet18'}
运行实验时，实验管理器使用表中指定的超参数值的每种组合来训练网络。
要通过使用贝叶斯优化找到最佳的培训选项，请设置StrategytoBayesian Optimization。在里面hyperparameter table, specify these properties of the hyperparameters used in the experiment:
- Name— Enter a valid hyperparameter name. Hyperparameter names must start with a letter, followed by letters, digits, or underscores.
- 范围— For a real- or integer-valued hyperparameter, enter a two-element vector that gives the lower bound and upper bound of the hyperparameter. For a categorical hyperparameter, enter an array of strings or a cell array of character vectors that lists the possible values of the hyperparameter.
- Type— Select真实的（实用值超参数），整数(integer-valued hyperparameter), orcategorical（分类超参数）。
- 转换— Selectnone（无转换）或log（对数变换）。为了log，，，，the hyperparameter must be真实的或者整数和positive. With this option, the hyperparameter is searched and modeled on a logarithmic scale.
When you run the experiment, Experiment Manager searches for the best combination of hyperparameters. Each trial in the experiment uses a new combination of hyperparameter values based on the results of the previous trials.
为了指定实验的持续时间，Bayesian Optimization Options，输入最长时间（以秒为单位）和运行的最大试验次数。请注意，实验中的实际运行时间和试验数可以超过这些设置，因为仅当试验完成执行时，实验管理器才会检查这些选项。
Bayesian optimization requires Statistics and Machine Learning Toolbox. For more information, seeTune Experiment Hyperparameters by Using Bayesian Optimization。

这Setup Function为实验配置培训数据，网络体系结构和培训选项。设置函数的输入是一个结构，该结构具有来自超参数表的字段。设置功能的输出必须与trainNetwork功能。This table lists the supported signatures for the setup function.

Goal of Experiment	Setup Function Signature
Train a network for image classification and regression tasks using the images and responses specified by`images`和the training options defined by`options`。	功能[图像，层，选项] = perveriment_setup（params）。。。end
Train a network using the images specified by`images`和由`responses`。	功能(images,responses,layers,options] = Experiment_setup(params)。。。end
Train a network for sequence or time-series classification and regression tasks (for example, an LSTM or GRU network) using the sequences and responses specified by`sequences`。	功能[序列，层，选项] = perveriment_setup（params）。。。end
使用由`sequences`和由`responses`。	功能(sequences,reponses,layers,options] = Experiment_setup(params)。。。end
训练网络进行特征分类或回归任务（例如，使用由特征数据和响应`features`。	功能[功能，层，选项] = perveriment_setup（params）。。。end
使用指定的功能数据训练网络`features`和由`responses`。	功能(features,responses,layers,options] = Experiment_setup(params)。。。end

小费

When writing your setup function, follow these guidelines:

加载培训和验证数据通过使用绝对路径到达所有工人都可以访问的位置。
对于包含批准层的网络，如果BatchNormalizationStatisticstraining option is人口，实验管理器显示最终验证度量值，这些度量值通常与培训期间评估的验证指标不同。值的差异是网络完成培训后执行的其他操作的结果。有关更多信息，请参阅Batch Normalization Layer。

这execution modes that you can use for your experiment depend on the settings you specify for the training optionsExecutionEnvironment和DispatchInBackground。

Execution Mode	Valid Settings for`ExecutionEnvironment`	Valid Settings for`DispatchInBackground`
`顺序`	`"auto"`，，，，`“中央处理器”`，，，，`"gpu"`，，，，`"multi-gpu"`，，，，`"parallel"`	`真的`，，，，`false`
`同时`	`"auto"`，，，，`“中央处理器”`，，，，`"gpu"`	`false`
`Batch Sequential`	`"auto"`，，，，`“中央处理器”`，，，，`"gpu"`，，，，`"parallel"`	`真的`，，，，`false`
`批量同时`	`"auto"`，，，，`“中央处理器”`，，，，`"gpu"`	`false`

有关更多信息，请参阅使用实验经理并行培训网络和Offload Experiments as Batch Jobs to Cluster。

这指标section specifies functions to evaluate the results of the experiment. The input to a metric function is a structure with three fields:

trainedNetwork是个系列网络对象或DAGNetwork目的returned by thetrainNetwork功能。有关更多信息，请参阅训练有素的网络。
训练is a structure containing the training information returned by thetrainNetwork功能。有关更多信息，请参阅培训信息。
参数is a structure with fields from the hyperparameter table.

度量函数的输出必须是标量号，逻辑值或字符串。

If your experiment uses Bayesian optimization, select a metric to optimize from the优化list. In the方向list, specify that you want toMaximize或者最小化this metric. Experiment Manager uses this metric to determine the best combination of hyperparameters for your experiment. You can choose a standard training or validation metric (such as accuracy, RMSE, or loss) or a custom metric from the table.

Configure Custom Training Experiment

This example shows how to set up a custom training experiment using the Experiment Manager app. Custom training experiments support workflows that require a training function other thantrainNetwork。这些工作流程包括：

训练未由图层图定义的网络。
Training a network using a custom learning rate schedule.
通过使用自定义功能更新网络的可学习参数。
Training a generative adversarial network (GAN).
Training a Siamese network.

Custom training experiments consist of a description, a table of hyperparameters, and a training function.

在里面Descriptionfield, enter a description of the experiment.

UnderHyperparameters，，，，select the strategy to use for your experiment.

To sweep through a range of hyperparameter values, setStrategytoExhaustive Sweep。在里面hyperparameter table, enter the names and values of the hyperparameters used in the experiment. Hyperparameter names must start with a letter, followed by letters, digits, or underscores. Hyperparameter values must be scalars or vectors with numeric, logical, or string values, or cell arrays of character vectors. For example, these are valid hyperparameter specifications:
- 0.01
- 0.01：0.01：0.05
- (0.010.02 0.04 0.08]
- ("sgdm" "rmsprop" "adam"]
- {'squeezenet' 'googlenet' 'resnet18'}
运行实验时，实验管理器使用表中指定的超参数值的每种组合来训练网络。
要通过使用贝叶斯优化找到最佳的培训选项，请设置StrategytoBayesian Optimization。在里面hyperparameter table, specify these properties of the hyperparameters used in the experiment:
- Name— Enter a valid hyperparameter name. Hyperparameter names must start with a letter, followed by letters, digits, or underscores.
- 范围— For a real- or integer-valued hyperparameter, enter a two-element vector that gives the lower bound and upper bound of the hyperparameter. For a categorical hyperparameter, enter an array of strings or a cell array of character vectors that lists the possible values of the hyperparameter.
- Type— Select真实的（实用值超参数），整数(integer-valued hyperparameter), orcategorical（分类超参数）。
- 转换— Selectnone（无转换）或log（对数变换）。为了log，，，，the hyperparameter must be真实的或者整数和positive. With this option, the hyperparameter is searched and modeled on a logarithmic scale.
When you run the experiment, Experiment Manager searches for the best combination of hyperparameters. Each trial in the experiment uses a new combination of hyperparameter values based on the results of the previous trials.
为了指定实验的持续时间，Bayesian Optimization Options，输入最长时间（以秒为单位）和运行的最大试验次数。请注意，实验中的实际运行时间和试验数可以超过这些设置，因为仅当试验完成执行时，实验管理器才会检查这些选项。
Bayesian optimization requires Statistics and Machine Learning Toolbox. For more information, seeUse Bayesian Optimization in Custom Training Experiments。

这Training Functionspecifies the training data, network architecture, training options, and training procedure used by the experiment. The inputs to the training function are:

从hyperparameter ta结构字段ble
一个experiments.Monitor您可以使用的对象来跟踪培训的进度，更新结果表中的信息字段，记录培训指标的值以及生产培训图

实验管理器保存了培训功能的输出，因此您可以在培训完成后将其导出到MATLAB工作区。

小费

编写培训功能时，请遵循以下准则：

加载培训和验证数据通过使用绝对路径到达所有工人都可以访问的位置。
Both information and metric columns display numerical values in the results table for your experiment. Additionally, metric values are recorded in the training plot. Use information columns for values that you want to display in the results table but not in the training plot.

If your experiment uses Bayesian optimization, in the指标section, under优化，输入以优化的度量标准的名称。在里面方向list, specify that you want toMaximize或者最小化this metric. Experiment Manager uses this metric to determine the best combination of hyperparameters for your experiment. You can choose any metric that you define using theexperiments.Monitor训练功能的对象。

停止，，，，Cancel, and Restart Training

实验经理provides two options for interrupting experiments:

停止将任何运行试验标记为停止ped和saves their results. When the experiment stops, you can display the training plot and export the training output for these trials.
Cancel将任何运行试验标记为取消和discards their results. When the experiment stops, you cannot display the training plot or export the training output for these trials.

Both options save the results of any completed trials and cancel any queued trials. Typically,Cancel比快停止。

您可以停止正在运行或取消单个排队试验的个人试验，而不是停止实验。在里面Actionscolumn of the results table, click the Stop button或取消按钮for the trial.

Results table showing Stop icon for a running trial.

When the training is complete, you can restart a trial that you stopped or canceled. In theActions结果表的列，单击重新启动按钮for the trial.

Results table showing Restart icon for a running trial.

Alternatively, to restart all the trials that you canceled, in the Experiment Manager toolstrip, clickRestart All Canceled。

笔记

所有实验类型，策略或执行模式都不可用停止，取消和重新启动选项。

Sort, Filter, and Annotate Experiment Results

This example shows how to compare your results and record your observations after running an experiment.

When you run an experiment, Experiment Manager trains the network defined by the setup function multiple times. Each trial uses a different combination of hyperparameters. When the experiment finishes, a table displays training and validation metrics (such as RMSE and loss) for each trial. To compare the results of an experiment, you can use these metrics to sort the results table and filter trials.

要对结果表中的试验进行排序，请使用对应于培训或验证度量的列的下拉列表。

指向您要排序的列的标题。
Click the triangle icon.
SelectSort in Ascending Order或者降序。

To filter trials from the results table, use theFilters窗格。This pane shows histograms for the numeric metrics in the results table. To remove a histogram from theFilterspane, in the results table, open the drop-down list for the corresponding column and clear the显示过滤器check box.

在实验经理工具条，选择Filters。
调整直方图下的滑块，以进行要过滤的训练或验证度量。

结果表仅显示在选定范围内具有度量值的试验。
To restore all of the trials in the results table, close the实验结果pane and reopen the results from the实验浏览器窗格。

To record observations about the results of your experiment, add an annotation.

Right-click a cell in the results table and selectAdd Annotation。Alternatively, select a cell in the results table and, on the Experiment Manager toolstrip, select一个notations>Add Annotation。
在里面一个notationspane, enter your observations in the text box. You can add multiple annotations for each cell in the results table.
要突出显示与注释相对应的单元格，请单击注释上方的链接。

To sort annotations, use the排序方式drop-down list. You can sort by creation time or trial number.

Sorting options include: Creation Time: Older to Newer (Default), Creation Time: Newer to Older, Trial Number: Lowest to Highest, and Trial Number: Highest to Lowest.

To delete an annotation, click the delete button在注释的右边。

View Source of Past Experiment Definitions

此示例显示了如何检查产生给定结果的实验的配置。

After you run an experiment, you can open the实验Sourcepane to see a read-only copy of the experiment description and hyperparameter table, as well as links to all of the functions used by the experiment. You can use the information in this pane to track the configuration of data, network, and training options that produces each of your results.

为了instance, suppose that you run an experiment multiple times. Each time that you run the experiment, you change the contents of the setup function but always use the same function name. The first time that you run the experiment, you use the default network provided by the experiment template for image classification. The second time that you run the experiment, you modify the setup function to load a pretrained GoogLeNet network, replacing the final layers with new layers for transfer learning. For an example that uses these two network architectures, seeCreate a Deep Learning Experiment for Classification。

在第一个实验结果pane, click theView Experiment Sourcelink. Experiment Manager opens an实验Sourcepane that contains the experiment definition that produced the first set of results. Click the link at the bottom of the pane to open the setup function that you used the first time you ran the experiment. You can copy this setup function to rerun the experiment using the simple classification network.

在second实验结果pane, click theView Experiment Sourcelink. Experiment Manager opens an实验Source包含产生第二组结果的实验定义的窗格。单击窗格底部的链接以打开您第二次运行实验时使用的设置功能。您可以使用转移学习复制此设置功能以重新运行实验。

实验管理器存储您使用的所有功能的副本，因此您不必在修改和重新运行实验时手动重命名这些功能。

小费s

为了可视化，构建和训练网络而无需大量的超参数，您可以使用Deep Network Designer应用程序。训练网络后，生成一个实验以找到最佳的培训选项。有关更多信息，请参阅Generate Experiment Using Deep Network Designer。
To run an experiment in parallel using MATLAB Online, you must have access to a Cloud Center cluster. For more information, see在MATLAB在线使用并行计算工具箱与Cloud Center群集(Parallel Computing Toolbox)。
To navigate Experiment Manager when using a mouse is not an option, use shortcut keyboards. For more information, seeKeyboard Shortcuts for Experiment Manager。

实验经理

Description

More

打开实验经理应用程序

Examples

Image Classification by Sweeping Hyperparameters

图像回归通过清除超参数

Image Classification Using Custom Training Loop

Configure Built-In Training Experiment

Configure Custom Training Experiment

停止，，，，Cancel, and Restart Training

Sort, Filter, and Annotate Experiment Results

View Source of Past Experiment Definitions

相关示例

小费s

版本历史记录

也可以看看

Apps

功能

Objects

Topics