Note
Go to the end to download the full example code. or to run this example in your browser via Binder
FREM on Jimura et al “mixed gambles” dataset¶
In this example, we use fast ensembling of regularized models (FREM) to solve a regression problem, predicting the gain level corresponding to each beta maps regressed from mixed gambles experiment. FREM uses an implicit spatial regularization through fast clustering and aggregates a high number of estimators trained on various splits of the training set, thus returning a very robust decoder at a lower computational cost than other spatially regularized methods.
To have more details, see: FREM: fast ensembling of regularized models for robust decoding.
Load the data from the Jimura mixed-gamble experiment¶
from nilearn.datasets import fetch_mixed_gambles
data = fetch_mixed_gambles(n_subjects=16)
zmap_filenames = data.zmaps
behavioral_target = data.gain
mask_filename = data.mask_img
[get_dataset_dir] Dataset found in /home/runner/work/nilearn/nilearn/nilearn_data/jimura_poldrack_2012_zmaps
Fit FREM¶
We compare both of these models to a pipeline ensembling many models
from nilearn.decoding import FREMRegressor
frem = FREMRegressor("svr", cv=10, standardize="zscore_sample")
frem.fit(zmap_filenames, behavioral_target)
# Visualize FREM weights
# ----------------------
from nilearn.plotting import plot_stat_map
plot_stat_map(
frem.coef_img_["beta"],
title="FREM",
display_mode="yz",
cut_coords=[20, -2],
threshold=0.2,
)
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/nilearn/decoding/decoder.py:746: UserWarning:
Brain mask is bigger than the volume of a standard human brain. This object is probably not tuned to be used on such data.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
/opt/hostedtoolcache/Python/3.12.6/x64/lib/python3.12/site-packages/sklearn/svm/_base.py:297: ConvergenceWarning:
Solver terminated early (max_iter=10000). Consider pre-processing your data with StandardScaler or MinMaxScaler.
<nilearn.plotting.displays._slicers.YZSlicer object at 0x7f42916bf5f0>
We can observe that the coefficients map learnt by FREM is structured, due to the spatial regularity imposed by working on clusters and model ensembling. Although these maps have been thresholded for display, they are not sparse (i.e. almost all voxels have non-zero coefficients). See also this other example using FREM, and related section of user guide.
Example use of TV-L1 SpaceNet¶
SpaceNet is another method available in Nilearn to decode with spatially sparse models. Depending on the penalty that is used, it yields either very structured maps (TV-L1) or unstructured maps (graph_net). Because of their heavy computational costs, these methods are not demonstrated on this example but you can try them easily if you have a few minutes. Example code is included below.
from nilearn.decoding import SpaceNetRegressor
# We use the regressor object since the task is to predict a continuous
# variable (gain of the gamble).
tv_l1 = SpaceNetRegressor(
mask=mask_filename,
penalty="tv-l1",
eps=1e-1, # prefer large alphas
memory="nilearn_cache",
n_jobs=2,
)
# tv_l1.fit(zmap_filenames, behavioral_target)
# plot_stat_map(tv_l1.coef_img_, title="TV-L1", display_mode="yz",
# cut_coords=[20, -2])
Total running time of the script: (0 minutes 24.167 seconds)
Estimated memory usage: 2453 MB