0

我在获取 GPyTorch 回归模型的 SHAP 值时遇到问题。

以下是我正在使用的 GPyTorch 代码：

import math
import torch
import gpytorch


'''
Data
'''

num_features = 3

# Training data is 100 points in [0,1] 
train_x = torch.rand(100, num_features)

# True function is sin(2*pi*x) with Gaussian noise
train_y = torch.sin(train_x[:,0] * (2 * math.pi)) + torch.randn(train_x[:,0].size()) * math.sqrt(0.04)

test_x = torch.rand(50, num_features)


'''
GP Model
'''

kernel = gpytorch.kernels.RBFKernel()    

# We will use the simplest form of GP model, exact inference
class ExactGPModel(gpytorch.models.ExactGP):
    def __init__(self, train_x, train_y, likelihood, kernel):
        super(ExactGPModel, self).__init__(train_x, train_y, likelihood)
        self.mean_module = gpytorch.means.ConstantMean()
        self.covar_module = gpytorch.kernels.ScaleKernel(kernel)   

    def forward(self, x):
        mean_x = self.mean_module(x)
        covar_x = self.covar_module(x)
        return gpytorch.distributions.MultivariateNormal(mean_x, covar_x)

# initialize likelihood and model
likelihood = gpytorch.likelihoods.GaussianLikelihood()
model = ExactGPModel(train_x, train_y, likelihood, kernel)
    

'''
Train
'''

training_iter = 50


# Find optimal model hyperparameters
# Put model on train mode
model.train()
likelihood.train()


# Use the adam optimizer
optimizer = torch.optim.Adam(model.parameters(), lr=0.1)  

# "Loss" for GPs - the marginal log likelihood
mll = gpytorch.mlls.ExactMarginalLogLikelihood(likelihood, model)

for i in range(training_iter):
    # Zero gradients from previous iteration
    optimizer.zero_grad()
    # Output from model
    output = model(train_x)
    # Calc loss and backprop gradients
    loss = -mll(output, train_y)
    loss.backward()
    print('Iter %d/%d - Loss: %.3f   lengthscale: %.3f   noise: %.3f' % (
        i + 1, training_iter, loss.item(),
        #model.covar_module.base_kernel.kernels[0].lengthscale.item(),
        model.covar_module.base_kernel.lengthscale.item(),
        model.likelihood.noise.item()   # ?
    ))
    optimizer.step()


'''
Prediction
'''

# Get into evaluation (predictive posterior) model
model.eval()
likelihood.eval()

with torch.no_grad(), gpytorch.settings.fast_pred_var():
    y_pred = likelihood(model(test_x))

y_pred_mean = y_pred.mean

我尝试了不同的方法从模型中获取 SHAP 值，但都没有成功。我尝试了什么：

"""
SHAP
"""
import shap

train_x_arr = train_x.detach().numpy() 
test_x_arr = test_x.detach().numpy()

第一次尝试：

# explain all the predictions in the test set
explainer = shap.KernelExplainer(model, train_x_arr)
shap_values = explainer.shap_values(test_x_arr)

错误

AttributeError: 'numpy.ndarray' object has no attribute 'ndimension'

第二次尝试：

explainer = shap.Explainer(model)
shap_values = explainer(test_x_arr)

错误：

TypeError: 'NoneType' object is not callable

如果您有任何解决此问题的想法，请告诉我。

4

0 回答 0