Reference documentation and code samples for the Google Cloud Ai Platform V1 Client class IntegratedGradientsAttribution.
An attribution method that computes the Aumann-Shapley value taking advantage of the model's fully differentiable structure. Refer to this paper for more details: https://arxiv.org/abs/1703.01365
Generated from protobuf message google.cloud.aiplatform.v1.IntegratedGradientsAttribution
Methods
__construct
Constructor.
| Parameters | |
|---|---|
| Name | Description |
data |
array
Optional. Data for populating the Message object. |
↳ step_count |
int
Required. The number of steps for approximating the path integral. A good value to start is 50 and gradually increase until the sum to diff property is within the desired error range. Valid range of its value is [1, 100], inclusively. |
↳ smooth_grad_config |
Google\Cloud\AIPlatform\V1\SmoothGradConfig
Config for SmoothGrad approximation of gradients. When enabled, the gradients are approximated by averaging the gradients from noisy samples in the vicinity of the inputs. Adding noise can help improve the computed gradients. Refer to this paper for more details: https://arxiv.org/pdf/1706.03825.pdf |
↳ blur_baseline_config |
Google\Cloud\AIPlatform\V1\BlurBaselineConfig
Config for IG with blur baseline. When enabled, a linear path from the maximally blurred image to the input image is created. Using a blurred baseline instead of zero (black image) is motivated by the BlurIG approach explained here: https://arxiv.org/abs/2004.03383 |
getStepCount
Required. The number of steps for approximating the path integral.
A good value to start is 50 and gradually increase until the sum to diff property is within the desired error range. Valid range of its value is [1, 100], inclusively.
| Returns | |
|---|---|
| Type | Description |
int | |
setStepCount
Required. The number of steps for approximating the path integral.
A good value to start is 50 and gradually increase until the sum to diff property is within the desired error range. Valid range of its value is [1, 100], inclusively.
| Parameter | |
|---|---|
| Name | Description |
var |
int
|
| Returns | |
|---|---|
| Type | Description |
$this | |
getSmoothGradConfig
Config for SmoothGrad approximation of gradients.
When enabled, the gradients are approximated by averaging the gradients from noisy samples in the vicinity of the inputs. Adding noise can help improve the computed gradients. Refer to this paper for more details: https://arxiv.org/pdf/1706.03825.pdf
| Returns | |
|---|---|
| Type | Description |
Google\Cloud\AIPlatform\V1\SmoothGradConfig|null | |
hasSmoothGradConfig
clearSmoothGradConfig
setSmoothGradConfig
Config for SmoothGrad approximation of gradients.
When enabled, the gradients are approximated by averaging the gradients from noisy samples in the vicinity of the inputs. Adding noise can help improve the computed gradients. Refer to this paper for more details: https://arxiv.org/pdf/1706.03825.pdf
| Parameter | |
|---|---|
| Name | Description |
var |
Google\Cloud\AIPlatform\V1\SmoothGradConfig
|
| Returns | |
|---|---|
| Type | Description |
$this | |
getBlurBaselineConfig
Config for IG with blur baseline.
When enabled, a linear path from the maximally blurred image to the input image is created. Using a blurred baseline instead of zero (black image) is motivated by the BlurIG approach explained here: https://arxiv.org/abs/2004.03383
| Returns | |
|---|---|
| Type | Description |
Google\Cloud\AIPlatform\V1\BlurBaselineConfig|null | |
hasBlurBaselineConfig
clearBlurBaselineConfig
setBlurBaselineConfig
Config for IG with blur baseline.
When enabled, a linear path from the maximally blurred image to the input image is created. Using a blurred baseline instead of zero (black image) is motivated by the BlurIG approach explained here: https://arxiv.org/abs/2004.03383
| Parameter | |
|---|---|
| Name | Description |
var |
Google\Cloud\AIPlatform\V1\BlurBaselineConfig
|
| Returns | |
|---|---|
| Type | Description |
$this | |