If you have a neuron model and you want to find the gabor filter that maximizes the activity of your model neuron, this tool might come handy - it finds the input arguments to a gabor-generating function via gradient descent.
Let's create a simple linear-nonlinear neuron model with a known receptive field:
import torch
from torch import nn
from torch.nn import functional as F
class Neuron(nn.Module):
def __init__(self, rf):
super().__init__()
self.rf = torch.tensor(rf.astype(np.float32))
def forward(self, x):
return F.elu((x * self.rf).sum()) + 1
from fitgabor.utils import gabor_fn
theta = -np.pi/4
groundtruth_rf = gabor_fn(theta, sigma=4, Lambda=14, psi=np.pi/2, gamma=1, center=(15, 5), size=(64, 64))
neuron = Neuron(groundtruth_rf)Here is the ground truth RF:
from fitgabor import GaborGenerator, trainer_fn
gabor_gen = GaborGenerator(image_size=(64, 64))
gabor_gen, _ = trainer_fn(gabor_gen, neuron)You can generate the learned gabor simply by calling the model:
learned_gabor = gabor_gen().data.numpy()And here is the learning evolution:
Let me know, please! I would love to know why it did not work and help fix it 👷
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent