Add files via upload

todo/optimal_stimulus.py

@@ -1,11 +1,23 @@
 import torch
 import matplotlib.pyplot as plt
+import matplotlib.patches as patch
 import matplotlib as mpl
+from functions.analyse_network import analyse_network
 
 mpl.rcParams["text.usetex"] = True
 mpl.rcParams["font.family"] = "serif"
 
 
+# define parameters
+num_iterations: int = 100000
+learning_rate: float = 0.1  # 0.0005  #
+apply_input_mask: bool = True
+mark_region_in_plot: bool = False
+sheduler_patience: int = 500
+sheduler_factor: float = 0.9
+sheduler_eps = 1e-08
+target_image_active: float = 1e4
+
 # path to NN
 nn = "network_0_seed0_Coignless_83Epoch_2807-1455"
 PATH = f"./trained_models/{nn}.pt"
@@ -30,6 +42,12 @@ input_img = torch.randn(1, 200, 200).to(device)
 input_img = input_img.unsqueeze(0)
 input_img.requires_grad_(True)  # type: ignore
 
+input_shape = input_img.shape
+assert input_shape[-2] == input_shape[-1]
+coordinate_list, layer_type_list, pixel_used = analyse_network(
+    model=model, input_shape=int(input_shape[-1])
+)
+
 output_shape = model(input_img).shape
 target_image = torch.zeros(
     (*output_shape,), dtype=input_img.dtype, device=input_img.device
@@ -37,9 +55,6 @@ target_image = torch.zeros(
 
 input_parameter = torch.nn.Parameter(input_img)
 
-# define parameters
-num_iterations: int = 10000
-learning_rate: float = 0.0005
 
 print(
     (
@@ -54,14 +69,43 @@ neuron_x = target_image.shape[2] // 2
 neuron_y = target_image.shape[3] // 2
 print(f"Selected neuron {neuron_f}, {neuron_x}, {neuron_y}")
 
+# Input mask ->
+active_input_x = coordinate_list[-1][:, neuron_x].clone()
+active_input_y = coordinate_list[-1][:, neuron_y].clone()
+
+input_mask: torch.Tensor = torch.zeros_like(input_img)
+
+input_mask[
+    :,
+    :,
+    active_input_x.type(torch.int64).unsqueeze(-1),
+    active_input_y.type(torch.int64).unsqueeze(0),
+] = 1
+
+rect_x = [int(active_input_x.min()), int(active_input_x.max())]
+rect_y = [int(active_input_y.min()), int(active_input_y.max())]
+# <- Input mask
+
+if apply_input_mask:
+    with torch.no_grad():
+        input_img *= input_mask
+
 
 optimizer = torch.optim.Adam([{"params": input_parameter}], lr=learning_rate)
-# TODO:
-# scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(optimizer)
 
-target_image[0, neuron_f, neuron_x, neuron_y] = 1e4
+scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(
+    optimizer,
+    patience=sheduler_patience,
+    factor=sheduler_factor,
+    eps=sheduler_eps * 0.1,
+)
 
-for i in range(num_iterations):
+target_image[0, neuron_f, neuron_x, neuron_y] = target_image_active
+
+backup_image = input_parameter.data.detach().clone()
+
+counter: int = 0
+while (optimizer.param_groups[0]["lr"] > sheduler_eps) and (counter < num_iterations):
     optimizer.zero_grad()
 
     output = model(input_parameter)
@@ -69,13 +113,46 @@ for i in range(num_iterations):
     loss = torch.nn.functional.mse_loss(output, target_image)
     loss.backward()
 
-    if i % 1000 == 0:
-        print(f"{i} : loss={float(loss):.3e} lr={optimizer.param_groups[0]['lr']:.3e}")
-    optimizer.step()
-    # TODO:
-    # scheduler.step(float(loss))
+    if counter % 1000 == 0:
+        print(
+            f"{counter} : loss={float(loss):.3e} lr={optimizer.param_groups[0]['lr']:.3e}"
+        )
 
+    optimizer.step()
+
+    if apply_input_mask:
+        with torch.no_grad():
+            input_parameter.data[torch.where(input_mask == 0)] = 0.0
+
+    if (
+        torch.isfinite(input_parameter.data).sum().cpu()
+        != torch.tensor(input_parameter.data.size()).prod()
+    ):
+        print(f"Found NaN in step: {counter}, use a smaller initial lr")
+        input_parameter.data = backup_image.clone()
+        exit()
+    else:
+        backup_image = input_parameter.data.detach().clone()
+
+    scheduler.step(float(loss))
+    counter += 1
 
 # plot image:
-plt.imshow(input_img.squeeze().detach().cpu().numpy(), cmap="gray")
+_, ax = plt.subplots()
+
+ax.imshow(input_img.squeeze().detach().cpu().numpy(), cmap="gray")
+
+if mark_region_in_plot:
+    edgecolor = "sienna"
+    kernel = patch.Rectangle(
+        (rect_y[0], rect_x[0]),
+        int(rect_y[1] - rect_y[0]),
+        int(rect_x[1] - rect_x[0]),
+        linewidth=1.2,
+        edgecolor=edgecolor,
+        facecolor="none",
+    )
+    ax.add_patch(kernel)
+
+
 plt.show(block=True)