Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Loading...
Get information about an anomaly block, such as its dependencies. Use the impulse blocks to find the learnId.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Selectable axes for the anomaly detection block
Whether the block is trained
Number of clusters for K-means, or number of components for GMM (in config)
Selected clusters (in config)
Minimum confidence rating for this block, scores above this number will be flagged as anomaly.
Download an exported Keras block - needs to be exported via 'exportKerasBlock' first
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
File
Download the data of an exported Keras block - needs to be exported via 'exportKerasBlockData' first
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
File
Download the processed data for this learning block. This is data already processed by the signal processing blocks.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
Numpy binary file
Download the labels for this learning block. This is data already processed by the signal processing blocks. Not all blocks support this function. If so, a GenericApiResponse is returned with an error message.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
Numpy binary file
Download a pretrained model file
Project ID
Impulse ID. If this is unset then the default impulse is used.
File
Download a trained model for a learning block. Depending on the block this can be a TensorFlow model, or the cluster centroids.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
Model download ID, which can be obtained from the project information
File
Configure the anomaly block, such as its minimum confidence score. Use the impulse blocks to find the learnId.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
Minimum confidence score, if the anomaly block scores a sample above this threshold it will be flagged as anomaly.
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Returns the latency, RAM and ROM used for the pretrained model - upload first via uploadPretrainedModel. This is using the project's selected latency device. Updates are streamed over the websocket API (or can be retrieved through the /stdout endpoint). Use getProfileTfliteJobResult to get the results when the job is completed.
Project ID
Impulse ID. If this is unset then the default impulse is used.
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Job identifier. Status updates will include this identifier.
12873488112Upload a pretrained model and receive info back about the input/output tensors. If you want to deploy a pretrained model from the API, see startDeployPretrainedModelJob.
Project ID
Impulse ID. If this is unset then the default impulse is used.
MCU used for calculating latency, query latencyDevices in listProject for a list of supported devices (and use the "mcu" property here). If this is kept empty then we'll show an overview of multiple devices.
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Job identifier. Status updates will include this identifier.
12873488112Get a sample of trained features, this extracts a number of samples and their features.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
Feature axis 1
Feature axis 2
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Total number of windows in the data set
Data by feature index for this window. Note that this data was scaled by the StandardScaler, use the anomaly metadata to unscale if needed.
Label used for datapoint colorscale in anomaly explorer (for gmm only). Is currently the result of the scoring function.
Get trained features for a single sample. This runs both the DSP prerequisites and the anomaly classifier.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
Sample ID
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Total number of windows in the data set
Data by feature index for this window. Note that this data was scaled by the StandardScaler, use the anomaly metadata to unscale if needed.
Label used for datapoint colorscale in anomaly explorer (for gmm only). Is currently the result of the scoring function.
Get raw model metadata of the Gaussian mixture model (GMM) for a trained anomaly block. Use the impulse blocks to find the learnId.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
2D array of shape (n, m)
3D array of shape (n, m, m)
1D array of shape (n,)
Save input / model configuration for a pretrained model. This overrides the current impulse. If you want to deploy a pretrained model from the API, see startDeployPretrainedModelJob.
Project ID
Impulse ID. If this is unset then the default impulse is used.
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Test out a pretrained model (using raw features) - upload first via uploadPretrainedModel. If you want to deploy a pretrained model from the API, see startDeployPretrainedModelJob.
Project ID
Impulse ID. If this is unset then the default impulse is used.
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Classification value per label. For a neural network this will be the confidence, for anomalies the anomaly score.
t-SNE2 output of the raw dataset using embeddings from this Keras block
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Data by feature index for this window
"`{ 0: 9.81, 11: 0.32, 22: 0.79 }`"Training label index
Training label string
Identifier for this block. Make sure to up this number when creating a new block, and don't re-use identifiers. If the block hasn't changed, keep the ID as-is. ID must be unique across the project and greather than zero (>0).
Block type (either time-series, image or features)
"time-series"Block name, will be used in menus
"Time series"Block title, used in the impulse UI
"Time series"Size of the sliding window in milliseconds
2004We use a sliding window to go over the raw data. How many milliseconds to increase the sliding window with for each step.
(Input only) Frequency of the input data in Hz
60We use a sliding window to go over the raw data. How many milliseconds to increase the sliding window with for each step in classification mode.
Whether to zero pad data when a data item is too short
Width all images are resized to before training
28Width all images are resized to before training
28How to resize images before training
"squash"Resize method to use when resizing images
"squash"If images are resized using a crop, choose where to anchor the crop
"middle-center"A short description of the block version, displayed in the block versioning UI
"Reduced learning rate and more layers"The system component that created the block version (createImpulse | clone | tuner). Cannot be set via API.
"createImpulse"The datetime that the block version was created. Cannot be set via API.
Only generate features for samples where (sample_id + datasetSubsetSeed) % datasetSubset) == 0
Get metadata about a trained anomaly block. Use the impulse blocks to find the learnId.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Date when the model was trained
Scale input for StandardScaler. Values are scaled like this (where ix is axis index): input[ix] = (input[ix] - mean[ix]) / scale[ix];
Mean input for StandardScaler. Values are scaled like this (where ix is axis index): input[ix] = (input[ix] - mean[ix]) / scale[ix];
Trained K-means clusters
Center of each cluster (one value per axis)
Size of the cluster
Which axes were included during training (by index)
"`[ 0, 11, 22 ]`"Default minimum confidence rating required before tagging as anomaly, based on scores of training data (GMM only).
The types of model that are available
Metrics for each of the available model types
The model's loss on the validation set after training
The model's accuracy on the validation set after training
Precision, recall, F1 and support scores
Custom, device-specific performance metrics
The name of the metric
The value of this metric for this model type
Only set for object detection projects
Only set for visual anomaly projects. 2D array of shape (n, n) with raw anomaly scores, where n varies based on the image input size and the specific visual anomaly algorithm used. The scores corresponds to each grid cell in the image's spatial matrix.
If this is set, then we're still profiling this model. Subscribe to job updates to see when it's done (afterward the metadata will be updated).
If this is set, then the profiling job failed (get the status by getting the job logs for 'profilingJobId').
Get metadata about a trained Keras block. Use the impulse blocks to find the learnId.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Date when the model was trained
Layers of the neural network
Input size
33TensorFlow name
"x_input:0"TensorFlow type
"<dtype: 'float32'>"Output size
20TensorFlow name
"dense_1/Relu:0"TensorFlow type
"<dtype: 'float32'>"Labels for the output layer
Original labels in the dataset when features were generated, e.g. used to render the feature explorer.
The types of model that are available
Metrics for each of the available model types
The model's loss on the validation set after training
The model's accuracy on the validation set after training
Precision, recall, F1 and support scores
Custom, device-specific performance metrics
The name of the metric
The value of this metric for this model type
Only set for object detection projects
Only set for visual anomaly projects. 2D array of shape (n, n) with raw anomaly scores, where n varies based on the image input size and the specific visual anomaly algorithm used. The scores corresponds to each grid cell in the image's spatial matrix.
If this is set, then we're still profiling this model. Subscribe to job updates to see when it's done (afterward the metadata will be updated).
If this is set, then the profiling job failed (get the status by getting the job logs for 'profilingJobId').
Normalization that is applied to images. If this is not set then 0..1 is used. "0..1" gives you non-normalized pixels between 0 and 1. "-1..1" gives you non-normalized pixels between -1 and 1. "0..255" gives you non-normalized pixels between 0 and 255. "-128..127" gives you non-normalized pixels between -128 and 127. "torch" first scales pixels between 0 and 1, then applies normalization using the ImageNet dataset (same as torchvision.transforms.Normalize()). "bgr-subtract-imagenet-mean" scales to 0..255, reorders pixels to BGR, and subtracts the ImageNet mean from each channel.
Receive info back about the earlier uploaded pretrained model (via uploadPretrainedModel) input/output tensors. If you want to deploy a pretrained model from the API, see startDeployPretrainedModelJob.
Project ID
Impulse ID. If this is unset then the default impulse is used.
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Whether a specific device was selected for performance profiling
The types of model that are available
Estimated amount of RAM required by the model, measured in bytes
Estimated amount of ROM required by the model, measured in bytes
Estimated arena size required for model inference, measured in bytes
Estimated amount of RAM required by the model, measured in bytes
Estimated amount of ROM required by the model, measured in bytes
Estimated arena size required for model inference, measured in bytes
Estimated amount of RAM required by the model, measured in bytes
Estimated amount of ROM required by the model, measured in bytes
Estimated arena size required for model inference, measured in bytes
Estimated amount of RAM required by the model, measured in bytes
Estimated amount of ROM required by the model, measured in bytes
Estimated arena size required for model inference, measured in bytes
Estimated amount of RAM required by the model, measured in bytes
Estimated amount of ROM required by the model, measured in bytes
Estimated arena size required for model inference, measured in bytes
Estimated amount of RAM required by the model, measured in bytes
Estimated amount of ROM required by the model, measured in bytes
Estimated arena size required for model inference, measured in bytes
Performance for a range of device types. Note that MPU is referred to as CPU in Studio, as MPU and CPU are treated equivalent for performance estimation.
Configure the Keras block, such as its minimum confidence score. Use the impulse blocks to find the learnId.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
Whether to use visual or expert mode.
Minimum confidence score, if the neural network scores a sample below this threshold it will be flagged as uncertain.
Raw Keras script (only used in expert mode)
The visual layers for the neural network (only in visual mode).
Number of neurons or filters in this layer (only for dense, conv1d, conv2d) or in the final conv2d layer (only for transfer layers)
Kernel size for the convolutional layers (only for conv1d, conv2d)
Fraction of input units to drop (only for dropout) or in the final layer dropout (only for transfer layers)
Number of columns for the reshape operation (only for reshape)
Number of convolutional layers before the pooling layer (only for conv1d, conv2d)
Custom transfer learning model ID (when type is set to transfer_organization)
Number of training cycles (only in visual mode).
Learning rate (between 0 and 1) (only in visual mode).
Batch size used during training (only in visual mode).
Train/test split (between 0 and 1)
Whether to automatically balance class weights, use this for skewed datasets.
Use learned optimizer and ignore learning rate.
The data augmentation policy to use with image input
True if spectrogram augmentation is enabled. Other properties will be ignored if this is false.
True if warping along the time axis is enabled.
The amount of frequency masking to apply.
The amount of time masking to apply.
The amount of Gaussian noise to add.
Whether to profile the i8 model (might take a very long time)
If set, skips creating embeddings and measuring memory (used in tests)
True if Akida Edge Learning model creation is enabled. Other properties will be ignored if this is false.
Number of additional classes that will be added to the Edge Learning model.
Number of neurons in each class on the last layer in the Edge Learning model.
If the 'custom validation split' experiment is enabled, this metadata key is used to prevent group data leakage between train and validation datasets.
Whether the 'Advanced training settings' UI element should be expanded.
Whether the 'Augmentation training settings' UI element should be expanded.
Training parameters, this list depends on the list of parameters that the model exposes.
Capacity level for visual anomaly detection. Determines which set of default configurations to use. The higher capacity, the higher number of (Gaussian) components, and the more adapted the model becomes to the original distribution
Training parameters specific to the type of the learn block. Parameters may be adjusted depending on the model defined in the visual layers. Used for our built-in blocks.
The backbone to use for feature extraction
The number of layers in the feature extractor (1-3)
The pool size for the feature extractor
The sampling ratio for the coreset, used for anomaly scoring
The number of nearest neighbors to consider, used for anomaly scoring
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Get information about a Keras block, such as its dependencies. Use the impulse blocks to find the learnId.
Project ID
Learn Block ID, use the impulse functions to retrieve the ID
OK
Whether the operation succeeded
Optional error description (set if 'success' was false)
Whether the block is trained
The type of learning block (anomaly, keras, keras-transfer-image, keras-transfer-kws, keras-object-detection, keras-regression). Each behaves differently.
The Keras script. This script might be empty if the mode is visual.
Minimum confidence rating required for the neural network. Scores below this confidence are tagged as uncertain.
The mode (visual or expert) to use for editing this network.
The visual layers (if in visual mode) for the neural network. This will be an empty array when in expert mode.
Number of neurons or filters in this layer (only for dense, conv1d, conv2d) or in the final conv2d layer (only for transfer layers)
Kernel size for the convolutional layers (only for conv1d, conv2d)
Fraction of input units to drop (only for dropout) or in the final layer dropout (only for transfer layers)
Number of columns for the reshape operation (only for reshape)
Number of convolutional layers before the pooling layer (only for conv1d, conv2d)
Custom transfer learning model ID (when type is set to transfer_organization)
Number of training cycles. If in expert mode this will be 0.
Learning rate (between 0 and 1). If in expert mode this will be 0.
The batch size used during training.
The default batch size if a value is not configured.
Python-formatted tuple of input axes
Train/test split (between 0 and 1)
Whether to automatically balance class weights, use this for skewed datasets.
Use learned optimizer and ignore learning rate.
The data augmentation policy to use with image input
True if spectrogram augmentation is enabled. Other properties will be ignored if this is false.
True if warping along the time axis is enabled.
The amount of frequency masking to apply.
The amount of time masking to apply.
The amount of Gaussian noise to add.
The type of learning block (anomaly, keras, keras-transfer-image, keras-transfer-kws, keras-object-detection, keras-regression). Each behaves differently.
URL to the source code of this custom learn block.
"Scale axes""text""Divide axes by this number""scale-axes"Interface section to render parameter in.
Only valid for type "string". Will render a multiline text area.
If set, shows a hint below the input.
Sets the placeholder text on the input element (for types "string", "int", "float" and "secret")
Category to display this block in the UI.
Whether to profile the i8 model (might take a very long time)
If set, skips creating embeddings and measuring memory (used in tests)
True if Akida Edge Learning model creation is enabled. Other properties will be ignored if this is false.
Number of additional classes that will be added to the Edge Learning model.
Number of neurons in each class on the last layer in the Edge Learning model.
This metadata key is used to prevent group data leakage between train and validation datasets.
Whether the 'Advanced training settings' UI element should be expanded.
Whether the 'Augmentation training settings' UI element should be expanded.
Training parameters, this list depends on the list of parameters that the model exposes.
Capacity level for visual anomaly detection. Determines which set of default configurations to use. The higher capacity, the higher number of (Gaussian) components, and the more adapted the model becomes to the original distribution
Training parameters specific to the type of the learn block. Parameters may be adjusted depending on the model defined in the visual layers. Used for our built-in blocks.
The backbone to use for feature extraction
The number of layers in the feature extractor (1-3)
The pool size for the feature extractor
The sampling ratio for the coreset, used for anomaly scoring
The number of nearest neighbors to consider, used for anomaly scoring