ディープラーニングツールのトレーニング

In Deep Learning Studio

1. Create a new tool.
2. Configure the tool settings.
3. Select the dataset split to use for this tool.
4. Configure the training settings and click on Train.

APIでは

In the API, to train a deep learning tool, call the corresponding method:

□ EClassifier.Train(trainingDataset, validationDataset, numberOfIterations)
□ ELocatorBase.Train(trainingDataset, validationDataset, numberOfIterations)
□ ESupervisedSegmenter.Train(trainingDataset, validationDataset, numberOfIterations)
□ EUnsupervisedSegmenter.Train(trainingDataset, validationDataset, numberOfIterations)

The training settings

● The Number of iterations. An Iteration corresponds to going through all the images in the training set once.
□ トレーニングプロセスは、良い結果を得るために多数のループを必要とします。
□ ループの数が多いほど、トレーニング時間は長くなり、得られる結果が向上します。

We recommend to use more iterations than the default value:
- For EasyClassify without pretraining, EasyLocate and EasySegment.
- For smaller dataset because the training is harder (for example, 100 iterations for a dataset with 100 images, 200 iterations for a dataset with 50 images, 400 iterations for a dataset with 25 images...).

□ By default, the number of iteration is determined automatically during the training to stop when no progress has been made for a while.

- This is called early stopping.

- To disable this behavior, use the method Train of any tool with the iteration parameter.

● The Batch size corresponds to the number of image patches that are processed together.
□ トレーニングはバッチのサイズの影響を受けます。
□ 大きなバッチのサイズは、GPU上の1つのループの処理速度を向上させますが、より多くのメモリを必要とします。
□ トレーニングプロセスでは、小さすぎるバッチのサイズでは良いモデルを学習できません。
□ デフォルトでは、バッチサイズは利用可能なメモリの観点からトレーニングのスピードを最適化するためにトレーニング中に自動的に決定されます。

- Use EDeepLearningTool.OptimizeBatchSize(false) to disable this behavior.

- Use EDeepLearningTool.BatchSize to change the size of your batch.

□ EDeepLearningTool.BatchSizeForMaximumInferenceSpeed gets the batch size that maximizes the batch classification speed on a GPU according to the available memory.
□ パフォーマンス上の理由から、バッチのサイズとして2のべき乗を選択するのが一般的です。
● The Data augmentation.
● Whether to use Deterministic training or not.
□ The deterministic training allows to reproduce the exact same results when all the settings are the same (tool settings, dataset split and training settings).
□ The deterministic training fixes random seeds used in the training algorithm and uses deterministic algorithms.
□ The deterministic training is usually slower than a non-deterministic training.
□ In Deep Learning Studio, the option to use deterministic training and the random seed are available in the advanced parameters.

□ In the API, use EDeepLearningTool.EnableDeterministicTraining. and EDeepLearningTool.DeterministicTrainingRandomSeed.

Continue the training

You can continue to train a tool that is already trained.

If you used the early stopping (the default behavior) for the previous training, provide the iteration parameter to continue the training.

In Deep Learning Studio, the dataset split associated with a trained tool is locked.

□ You can only continue training a tool with the same dataset split.
□ You can still add new training or validation images to the split by moving test images to the training set or the validation set of that split.

If you used the early stopping (the default behavior) for the previous training, uncheck the button Automatic and provide the number of iterations to continue the training.

Asynchronous training

The training process is asynchronous and performed in the background.

● Deep Learning Studio：
□ The training processes are queued.
□ They are automatically executed one after the other.
□ You can manually reorder the training in the processing queue.

● APIでは：
□ The Train method corresponding to your tool launches a new thread that does the training in the background.
□ EDeepLearningTool.WaitForTrainingCompletion suspends the program until the whole training is completed.
□ EDeepLearningTool.WaitForIterationCompletion suspends the program until the current iteration is completed.
□ During the training, EDeepLearningTool.CurrentTrainingProgression shows the progression of the training.