Commit c2600c21 authored by Ines Filipa Fernandes Ramos's avatar Ines Filipa Fernandes Ramos
Browse files

Update notebook_MEI_gen.ipynb

parent b14acfc2
%% Cell type:markdown id: tags:
# Notebook to generate MEIs for neurons of interest
%% Cell type:code id: tags:
``` python
%matplotlib inline
%load_ext autoreload
%load_ext memory_profiler
import torch
import matplotlib.pyplot as plt
from neuralpredictors.data.datasets import FileTreeDataset
import MEI
import matplotlib as mpl
import os
from os.path import join
import numpy as np
```
%%%% Output: stream
The autoreload extension is already loaded. To reload it, use:
%reload_ext autoreload
The memory_profiler extension is already loaded. To reload it, use:
%reload_ext memory_profiler
%% Cell type:markdown id: tags:
# Upload dataset
%% Cell type:code id: tags:
``` python
#Dataset directory
path_dataset = [r'C:\Users\inoca\Desktop\Intership EMBL\Python docs\Inception_loop\Ephy_data_Cortex\static20210629_train']
dataset_name = path_dataset[0].split("\\")[-1].replace('static','')
#Configure dataloader/s
from lurz2020.datasets.mouse_loaders import static_loaders
dataset_config = {'paths': path_dataset,
'batch_size': 64,
'seed': 1,
'cuda': True,
'normalize': True,
'include_eye_position': False,
'exclude': "images"}
dataloaders = static_loaders(**dataset_config)
dat = FileTreeDataset(path_dataset[0], "images", "responses")
```
%% Cell type:markdown id: tags:
# Upload trained models for MEI generation
%% Cell type:code id: tags:
``` python
#Load best tunned model instances
from lurz2020.models.models import se2d_fullgaussian2d
#Select number of instances to be used
n_seeds = 4
#Models name
models_name = 'tunned_model_ephy_data_cortex_'
#Directory where the models are saved
path_models = join(path_dataset[0], 'models')
#Build model object to load weights of trained model instances
model_config_tunned = {'init_mu_range': 0.1,
'init_sigma': 0.64,
'input_kern': 15,
'hidden_kern': 13,
'gamma_input': 1.0,
'grid_mean_predictor': None,
'gamma_readout': 0.99}
models = []
for n in range(n_seeds):
#Build model object to load weights of trained model instances
model = se2d_fullgaussian2d(**model_config_tunned, dataloaders=dataloaders, seed=n)
#Load weights from saved model state
model_state = torch.load(join(path_models,models_name+str(n)+'.pth'))
model.load_state_dict(model_state, strict=False)
if torch.cuda.is_available():
model.cuda()
#Change model mode to evaluation mode
model.eval()
models.append(model)
```
%% Cell type:code id: tags:
``` python
model.core
```
%%%% Output: execute_result
SE2dCore(
(_input_weights_regularizer): LaplaceL2norm(
(laplace): Laplace()
)
(features): Sequential(
(layer0): Sequential(
(conv): Conv2d(1, 64, kernel_size=(15, 15), stride=(1, 1), bias=False)
(norm): BatchNorm2d(64, eps=1e-05, momentum=0.9, affine=True, track_running_stats=True)
(nonlin): ELU(alpha=1.0, inplace=True)
)
(layer1): Sequential(
(ds_conv): DepthSeparableConv2d(
(in_depth_conv): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(spatial_conv): Conv2d(64, 64, kernel_size=(13, 13), stride=(1, 1), padding=(6, 6), groups=64, bias=False)
(out_depth_conv): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
)
(norm): BatchNorm2d(64, eps=1e-05, momentum=0.9, affine=True, track_running_stats=True)
(nonlin): ELU(alpha=1.0, inplace=True)
)
(layer2): Sequential(
(ds_conv): DepthSeparableConv2d(
(in_depth_conv): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(spatial_conv): Conv2d(64, 64, kernel_size=(13, 13), stride=(1, 1), padding=(6, 6), groups=64, bias=False)
(out_depth_conv): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
)
(norm): BatchNorm2d(64, eps=1e-05, momentum=0.9, affine=True, track_running_stats=True)
(nonlin): ELU(alpha=1.0, inplace=True)
)
(layer3): Sequential(
(ds_conv): DepthSeparableConv2d(
(in_depth_conv): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(spatial_conv): Conv2d(64, 64, kernel_size=(13, 13), stride=(1, 1), padding=(6, 6), groups=64, bias=False)
(out_depth_conv): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
)
(norm): BatchNorm2d(64, eps=1e-05, momentum=0.9, affine=True, track_running_stats=True)
(nonlin): ELU(alpha=1.0, inplace=True)
)
)
) [SE2dCore regularizers: gamma_input = 1.0|skip = 0]
%% Cell type:markdown id: tags:
# MEI generation
%% Cell type:code id: tags:
``` python
#Parameters for MEI generation
MEIParameter = {
#Parameters used for MEI generation for V1 neurons used in paper Walker et al. 2019:
#1000, 1.5, 0.01, 3.0, 0.125, 0.1, 0.1, 0, False, -1, 11.0)
#1000, 1.5, 0.01, 3.0, 0.125, 0.1, 0.1, 0, True, -1, -1)
'iter_n' : 1000, # int number of iterations to run
'start_sigma' : 2.34, # float starting sigma value
'end_sigma' : 0.16, # float ending sigma value
'start_step_size' : 3.0, # float starting step size
'end_step_size' : 0.125, # float ending step size
'precond' : 0.13, # float strength of gradient preconditioning filter falloff
'step_gain' : 0.1, # float scaling of gradient steps
'jitter' : 0, # int size of translational jittering
'blur' : True, # bool whether to apply bluring or not
'norm' : -1, # float norm adjustment after step, negative to turn off
'train_norm' : -1 # float norm adjustment during step, negative to turn off
}
```
%% Cell type:code id: tags:
``` python
#Initialize MEI class object by defining model/s, dataset and MEI algorithm parameters to be used
MEIS = MEI.multi_MEI_class(dataset_name = dataset_name, dat = dat, dataloaders = dataloaders, models = models, n_seeds = n_seeds, MEIParameter = MEIParameter)
```
%% Cell type:code id: tags:
``` python
#Select target neurons for MEI generation
targets_list = range(3)
for target in targets_list:
#Generate MEI for one target unit
mei_target = MEIS.generate(target, track = False) #track=True to plot neuron activation values and corresponding generated images through the gradient ascent algorithm
```
%%%% Output: stream
Working on neuron_id=0
Working with images with mu=127.67724537037037, sigma=55.45213271323961
getting image size:
starting drawing
finished step 0 in octave 0
finished step 100 in octave 0
finished step 200 in octave 0
finished step 300 in octave 0
finished step 400 in octave 0
finished step 500 in octave 0
finished step 600 in octave 0
finished step 700 in octave 0
finished step 800 in octave 0
finished step 900 in octave 0
%%%% Output: stream
100%|█████████████████████████████████████████████████████████████████████████████| 1000/1000 [00:05<00:00, 179.81it/s]
%%%% Output: stream
Working on neuron_id=1
Working with images with mu=127.67724537037037, sigma=55.45213271323961
getting image size:
starting drawing
finished step 0 in octave 0
finished step 100 in octave 0
finished step 200 in octave 0
finished step 300 in octave 0
finished step 400 in octave 0
finished step 500 in octave 0
finished step 600 in octave 0
finished step 700 in octave 0
finished step 800 in octave 0
finished step 900 in octave 0
%%%% Output: stream
100%|█████████████████████████████████████████████████████████████████████████████| 1000/1000 [00:05<00:00, 193.11it/s]
%%%% Output: stream
Working on neuron_id=2
Working with images with mu=127.67724537037037, sigma=55.45213271323961
getting image size:
starting drawing
finished step 0 in octave 0
finished step 100 in octave 0
finished step 200 in octave 0
finished step 300 in octave 0
finished step 400 in octave 0
finished step 500 in octave 0
finished step 600 in octave 0
finished step 700 in octave 0
finished step 800 in octave 0
finished step 900 in octave 0
%%%% Output: stream
100%|█████████████████████████████████████████████████████████████████████████████| 1000/1000 [00:05<00:00, 192.53it/s]
%% Cell type:code id: tags:
``` python
#Directory to save generated MEIs results
path_MEIs = join(path_dataset[0], "results")
os.makedirs(path_MEIs)
#Plot and save generated MEIs images
for TargetUnit in targets_list:
mei_target = MEIS.MEIProperties['neuron_id : '+str(TargetUnit)]['mei']
fig, axs = plt.subplots(1, figsize=(15,4))
axs.imshow(mei_target)
axs.set_title('MEI unit '+str(TargetUnit))
plt.savefig(join(path_MEIs,"MEI_unit_"+str(TargetUnit)+models_name+".png"))
```
%%%% Output: display_data
![](data:image/png;base64,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)
%%%% Output: display_data
![](data:image/png;base64,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)
%%%% Output: display_data
![](data:image/png;base64,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)
%% Cell type:code id: tags:
``` python
# Save generated MEIS properties from the MEIS object to directory
import pickle
info_to_save_MEIS = {'dataset_name': MEIS.dataset_name,
'models': MEIS.models,
'n_seeds': MEIS.n_seeds,
'MEIParameter': MEIS.MEIParameter,
'MEIProperties': MEIS.MEIProperties}
with open(join(path_MEIs, "MEIS_" + models_name + ".pkl"), 'wb') as file:
pickle.dump(info_to_save_MEIS, file, pickle.HIGHEST_PROTOCOL)
```
%% Cell type:markdown id: tags:
--------------------------------
%% Cell type:code id: tags:
``` python
torch.cuda.empty_cache()
```
%% Cell type:markdown id: tags:
-----------------------------
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment